Advanced Generative AI Developer

27 May 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Picking a Bedrock Model for High-Volume RAG

A million LLM requests a day, peaking at thirty per second, split across US and EU customers, with a P99 first-token target under 1.5 seconds and real reasoning over retrieved context. Bedrock has seven model families and four ways to buy capacity. Most of the landscape falls away once you name what actually decides it, and the real trick is what you do *after* you've picked the model.

Read article 01 Jun 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

How to Build a Citations-Required RAG Over 50K Internal Documents

Fifty thousand internal documents, five gigabytes of text, weekly churn, a three-second latency budget, per-user access control, and a citation in every single answer. The RAG landscape on Bedrock is bigger than one product and the interesting part of the design is what falls away once you name the five things that actually decide it.

Read article 03 Jun 2026 · 32 min read

Exam Room · Advanced Generative AI Developer

Combining RAG and Fine-Tuning for a Legal Contract Assistant

A legal-tech team wants a contract review assistant that understands two hundred thousand past matters, speaks in the firm's voice with clause-by-section citations, and refuses anything off-domain. A hundred thousand dollars, three months. RAG, fine-tuning, and continued pre-training each solve a different half of that sentence; the interesting answer is which two to pick, not which one.

Read article 08 Jun 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Designing Short-Term and Long-Term Memory for a Bedrock Chat Assistant

A customer-support assistant where the average conversation runs fifteen turns before it resolves, and returning users pick up two weeks later expecting the bot to remember they've been waiting on a refund. Two memory problems in one product (what's live in the current conversation and what persists across visits) and four plausible ways to build it. Managed memory on Bedrock AgentCore handles both halves; the trap is the half where teams reach for DynamoDB or a knowledge base and get it wrong.

Read article 15 Jun 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Configuring Bedrock Guardrails for PII, Topics, and Grounding

A consumer-facing chatbot on Bedrock has passed every red-team round on the obvious harms (no weapons, no hate, no CSAM) and is still shipping embarrassments: a card number pasted by one user echoing back in a reply, the bot comparing the company's product with a named competitor, and a hallucinated policy line that nobody in the building wrote. Five different filter jobs wrap the same Bedrock invocation, and Guardrails is the one surface that does all five without five Lambdas.

Read article 17 Jun 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Spreading Bedrock Load with Cross-Region Inference Profiles

A Bedrock-backed SaaS serving US, EU and APAC customers is hitting regional quota in us-east-1 during peak while the same model sits idle in eu-west-1. The team wants to spread load without fracturing the product into three regional deployments. Three letters on the front of the model ID do the job, provided the model supports it and the geography fits the customer.

Read article 19 Jun 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

Building RAG When the Source Documents Change Daily

A support assistant that has to answer from a product manual which the product team edits weekly, a pricing sheet that changes at month-end, and an operational runbook that mutates hourly. The base model doesn't know any of it, and fine-tuning won't keep up. Retrieval is the answer; the question is how much of the retrieval plumbing we want to own, and Bedrock Knowledge Bases, a LangChain stack, and a hand-rolled pipeline each put the lines in different places.

Read article 22 Jun 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

How to Wire an LLM to Side-Effecting Actions with Bedrock AgentCore

An assistant that has to look up a customer's subscription, pause it, refund a charge, and email confirmation. Not just answer, act. The glue between a language model and the rest of our systems comes down to one question: when the model asks to move money, what physically stops it until a human says yes? Bedrock AgentCore, a framework agent on our own infrastructure, and a hand-written tool router all put that gate in different places.

Read article 24 Jun 2026 · 23 min read

Exam Room · Advanced Generative AI Developer

Picking a Vector Store for Bedrock RAG

Twelve million embedding vectors, a 50ms retrieval budget, hybrid queries that mix keyword and semantic, and a bill that should not double the Bedrock spend on its own. OpenSearch Serverless, Aurora with pgvector, and Pinecone Serverless all serve the same shape of query, but their pricing curves, operational shapes, and query surfaces diverge the moment the corpus grows beyond demo scale.

Read article 29 Jun 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

How to Manage Prompts Across Thirty Services on Bedrock

One prompt scattered across thirty services, no versioning, no tests, drift between the copy in the code and the copy in the docs, a silent regression when somebody changed 'concise' to 'brief' and retention on one response tanked. Prompt engineering at a hundred callers isn't writing prose, it's configuration management. Bedrock Prompt Management, Git-backed templates, and parameterised prompts each solve a slice of the same problem.

Read article 01 Jul 2026 · 22 min read

Exam Room · Advanced Generative AI Developer

Evaluating LLM Output With Bedrock Eval Jobs

Two thousand historical support tickets, a summarisation prompt, a new model candidate, and a product manager asking whether switching would hurt quality. Bedrock evaluation jobs offer automated scoring, human review with a workforce you bring, and model comparison side by side, but they answer different questions, and getting the correct number out of the correct job matters more than running more of them.

Read article 03 Jul 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

How to Build a Multi-Modal Bedrock Assistant for Insurance Claims

A claims-processing assistant that reads a scanned invoice, listens to a voicemail, answers the customer's question in plain text, and, if asked, reads it back. Four modalities, one conversation. The model choice, the orchestration shape, and the ways different inputs fail each push the architecture in different directions, and the naive 'just use a multi-modal model' misses half of where the real work is.

Read article 06 Jul 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

How to Cut a Bedrock Bill Without Hurting Quality

A Bedrock bill that doubled in two months, a product team that blames the retrieval service, a finance partner who'd like a straight answer. The cheapest token is the one you don't send; the next cheapest is the one you send to the correct model. Model routing, prompt compression, cached retrieval, and provisioned throughput each solve a different slice of the cost problem, and none of them is the silver bullet.

Read article 08 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Importing Custom Weights into Bedrock

A research team has fine-tuned an open-weights model for medical-notes summarisation on a private SageMaker cluster. The resulting weights live in S3; the production runtime needs Bedrock's ergonomics. Custom model import bridges that gap, but it only works for certain base architectures, comes with throughput minimums, and quietly changes the cost model compared to on-demand foundation models.

Read article 13 Jul 2026 · 23 min read

Exam Room · Advanced Generative AI Developer

Streaming Responses to Cut First-Token Latency

A chat interface where users wait four seconds for any response on long generations, abandon rate creeping up, product asking why we can't do the typing-animation thing that every other assistant does. Streaming isn't just a UX polish, it changes how the entire response path has to work, from the SDK call through API Gateway to the browser, and each hop has its own way of getting it wrong.

Read article 15 Jul 2026 · 21 min read

Exam Room · Advanced Generative AI Developer

When a Document Won't Fit the Context Window

A 400-page contract, a 200-page policy manual, and a legal team asking 'what clauses govern refund disputes across both?' A 200,000-token context window sounds like enough until you realise what goes in with the documents. Chunking, map-reduce, hierarchical summarisation, and sliding context windows each answer a different question, and getting the boundaries correct is most of the battle.

Read article 17 Jul 2026 · 22 min read

Exam Room · Advanced Generative AI Developer

Picking an Embedding Model for Retrieval

An index with 20 million chunks, queries that need to work in English, Spanish, Portuguese, and Japanese, and a budget that won't bear re-embedding every six months when someone decides the new model is better. The embedding model quietly caps what a retrieval system can ever do. Titan, Cohere, and a self-hosted model all trade different things, and the comparison is messier than the marketing suggests.

Read article 17 Jul 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

Making a Bedrock App Audit-Ready

A support assistant built on Bedrock has run cleanly for months, and now a compliance review has landed on it. The auditors want to know who invoked the model, what prompts and completions flowed through it, which model version answered, who approved the deployment, and proof that no customer PII leaked into the logs. The app works. It just can't answer a single one of those questions yet.

Read article 18 Jul 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

Checking a Bedrock Feature for Bias and Explainability

A Bedrock-powered feature drafts summaries of job applicants and triages people-facing cases, so its outputs shape decisions about real people, and it is up for a responsible-AI review before launch. The team already knows how to stop the model saying toxic or off-topic things; now they are being asked to show the feature is fair across groups and that any given output can be explained and justified. Content safety was the easy part; 'is it fair, and can you explain it?' is a different set of jobs.

Read article 20 Jul 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

How to Match Bedrock Pricing to Workload Rhythm

Standard on-demand Bedrock is priced per token, throttled per minute, and latency-variable in a way that product hates. The Reserved tier guarantees tokens-per-minute at a fixed monthly price, Priority charges more per request for faster responses, Flex discounts latency-tolerant traffic, and the Batch API halves the rate for anything that can wait. The right pick changes with the shape of the workload: a 24/7 assistant and a burst-heavy report generator land on opposite sides of the same line.

Read article 22 Jul 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

Keeping PII Out of LLM Prompts and Logs

A claims assistant that has to answer questions about a customer's claim while keeping the customer's name, address, policy number, and medical details out of training data, out of logs, and out of anything a subpoena could touch later. PII redaction isn't one knob, it's four or five, in different places, each covering a different leak path. Comprehend, Bedrock Guardrails, custom redaction, and the prompt itself each handle a different shape of the problem.

Read article 24 Jul 2026 · 24 min read

Exam Room · Advanced Generative AI Developer

Evaluating a RAG Pipeline End to End

A citations-required assistant over internal documents answers most questions well and a few of them wrong, and the team cannot tell why. Was the right passage never retrieved, or was it retrieved and then ignored? Those are two different failures with two different fixes, and an evaluation that reports one score for the whole pipeline hides which one you have. This walks through separating the retrieval half from the generation half so the regression harness points at the part that actually broke.

Read article 24 Jul 2026 · 23 min read

Exam Room · Advanced Generative AI Developer

Hybrid Search and Reranking for Bedrock RAG

A support assistant on Bedrock retrieves by pure semantic search and keeps whiffing on exact-term queries: product codes, error codes, part numbers, acronyms, proper names. The embedding of 'ERR-4021' sits close to a hundred other error codes, so a plausible-but-wrong chunk outranks the one document that actually names it. The fix is two levers, hybrid retrieval and a reranker, and knowing which one the precision problem actually needs.

Read article 25 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Prompt Engineering Techniques That Move the Needle

A dozen prompt-engineering tricks are supposed to improve output, but half of them cost tokens without moving quality and a couple quietly widen a security hole. Zero-shot, few-shot, chain-of-thought, ReAct, schema prompting, system framing, and delimiters each work on a specific task shape and hurt on the wrong one. The job is matching the technique to the task rather than stacking all of them and hoping.

Read article 25 Jul 2026 · 36 min read

Exam Room · Advanced Generative AI Developer

Choosing a Model From the Bedrock Catalogue

The Bedrock catalogue lists models from a dozen providers across text, images, video, and embeddings, and the reflex is to reach for the biggest, cleverest one for everything. That reflex triples the bill and slows every call, because most workloads never needed the flagship. The job is matching the model class to the shape of the work, then picking the smallest model that clears the quality bar.

Read article 25 Jul 2026 · 37 min read

Exam Room · Advanced Generative AI Developer

Choosing an Inference Option for a GenAI Workload

A summarisation feature, a nightly re-scoring job, and a customer-facing chat surface all need to serve a model, but they have nothing else in common. One needs low latency on steady traffic, one runs offline overnight on millions of records, and one spikes at odd hours and sits idle the rest of the time. Bedrock and SageMaker hosting each offer several ways to serve inference, and the right one falls out of the traffic shape and the cost model more than the model itself.

Read article 25 Jul 2026 · 33 min read

Exam Room · Advanced Generative AI Developer

Fine-Tuning, Continued Pre-Training, or Distillation

Prompting stopped closing the gap and retrieval only goes so far, so the model itself has to change. But the same word covers three training routes: fine-tuning on labelled pairs, continued pre-training on raw domain text, and distillation into a smaller student. They need different data, chase different goals, and cost wildly different amounts to run and to serve.

Read article 25 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Managing Prompts With Bedrock Prompt Management

The prompt is the product, and yet it usually lives as a string literal glued together in application code, edited by whoever last touched the deploy, with no version and no way to roll back a bad wording change. Amazon Bedrock Prompt Management makes the prompt a first-class resource with input variables, an attached model and inference configuration, and immutable versions an application references by identifier. The decision is whether that managed store beats over a prompt kept in your own source control, and it turns on how many services share the wording, who needs to edit it, and whether prompts feed Bedrock Flows.

Read article 25 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Prompt Caching Versus Response Caching on Bedrock

Two things called caching sit in front of a Bedrock app and they are not the same lever. Prompt caching still calls the model but you pay a fraction for a shared prefix; response caching may skip the model entirely and risks serving a stale or wrong-match answer. Knowing which repeated part you are actually paying for, the input prefix or the whole request, decides which one you reach for, and whether you stack both.

Read article 25 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Getting Documents Into a Bedrock Knowledge Base

A Bedrock Knowledge Base is only as good as what gets into it, and the ingestion pipeline is where most retrieval quality is won or lost. Documents pass through a source connector, a parser, a chunker, and an embedding model before they ever reach the vector index, and each stage has a choice with real cost and quality consequences. The job is picking the connector, the parser, and the sync mode that fit the documents you actually have.

Read article 25 Jul 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Choosing a Chunking Strategy for Bedrock Knowledge Bases

A mixed corpus is going into an Amazon Bedrock Knowledge Base: hundred-page PDFs, structured policy documents with numbered sections and tables, and a pile of short FAQ entries. The default fixed-size chunking retrieves badly because answers land across a chunk boundary or the chunks are so large they dilute relevance, and the team needs a chunking strategy that fits the shape of the documents rather than fighting it.

Read article 25 Jul 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Which AWS Store Can Do Vector Search

A team already runs OpenSearch for logs, Aurora for the transactional data, DynamoDB for the session store, and a Mongo-compatible cluster for the catalogue. Now a knowledge base needs vector search, and the first question is not which vector database to buy. It is which of the stores already humming in the account can do the job once the right feature is switched on, which one you enable differently than you expect, and which one looks eligible but is a trap.

Read article 26 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Picking the Right Tool to Check and Govern GenAI Data

Every night a knowledge base pulls in support tickets, product docs, and a database export, and every night some fraction of that arrives malformed, half-empty, duplicated, or carrying customer PII into a place it should never reach. The model is only as trustworthy as the corpus behind it, so the pipeline needs a gate: something that ingests the raw data, checks it against rules, and governs who can see what before any of it reaches an embedding. AWS offers at least five tools that each cover a piece of that job, and they are easy to mix up.

Read article 26 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Retrieval Over Structured Data With Text-to-SQL

A finance team wants natural-language answers to questions like total revenue by region last quarter, and the obvious move is to embed every row and reach for vector search. It falls apart the moment the answer is a sum instead of a sentence. The right pattern turns the question into SQL, runs it against a known schema, and returns the computed result. This walks the axes that decide when retrieval belongs in a query engine, not a vector index, and how to run generated SQL safely.

Read article 26 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Defending a Bedrock App Against Prompt Injection

A support assistant on Bedrock reads user questions, pulls answers from a knowledge base, and can raise refunds through a tool. Every one of those surfaces is a way in for an attacker who wants to override the instructions, jailbreak the guardrails, or turn the refund tool against the business. No single control stops all of it, so the real question is how the layers stack.

Read article 26 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Securing a Bedrock App: IAM, PrivateLink, and Keys

A production assistant on Amazon Bedrock has four separate ways to leak, and each one is a different control plane. Who is allowed to call the model, how the traffic reaches Bedrock, who holds the encryption keys, and where the data lives are four questions, not one. Getting one right does nothing for the other three; they are layers you stack, not options you pick between.

Read article 26 Jul 2026 · 33 min read

Exam Room · Advanced Generative AI Developer

Tuning Fine-Tuning: Epochs, Learning Rate, and Batch Size

A fine-tuning job on Amazon Bedrock exposes a short list of hyperparameters, and the two failure modes sit at opposite ends of the same dial: too little training and the model never picks up the behaviour you paid for, too much and it memorises the examples and forgets how to do everything else. Epochs, learning rate, and batch size are the knobs, and the training and validation loss curves are how you read whether you have turned them too far. The job is matching the amount of training to the amount of data, then trusting a held-out evaluation over the loss number.

Read article 26 Jul 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Orchestrating Multiple Bedrock Agents

A support assistant that once answered one kind of question now has to look up billing, check delivery windows, raise a ticket, and draft a reply, and each of those wants different tools and different judgement. One agent can reason its way through all of it, a supervisor agent can farm the pieces out to specialists, and a Flow can run the whole thing as a fixed pipeline. The trade is flexibility against predictability, cost, and latency, and the right shape depends on whether the model or the designer decides what happens next.

Read article 26 Jul 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Running Agents in Production With Bedrock AgentCore

A prototype agent built on an open-source framework works on a laptop, then has to become something real subscribers hit all day, with sessions that must not bleed into each other, memory that survives a reconnect, credentials that let it touch internal systems without over-granting, and traces you can read when a run goes wrong. Bedrock AgentCore is the set of building blocks for that operational layer, framework-agnostic and model-agnostic, so the question is which pieces you actually need and where the managed line should sit. The trade is control against operational burden, and it turns on how much of the runtime, memory, identity, and observability you want to own yourself.

Read article 26 Jul 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Buy or Build: Amazon Quick Versus a Custom RAG App

An enterprise wants a chat assistant over its own documents, and the choice is Amazon Quick off the shelf or a retrieval app built on Bedrock Knowledge Bases. Quick connects to the data sources, does the retrieval, and respects each user's permissions out of the box; a custom build hands you every knob and every operational bill. The deciding question is rarely which is more powerful, it is how bespoke the requirement is and whether per-user document access control has to be enforced.

Read article 27 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

When a Purpose-Built AI Service Beats a Foundation Model

Reaching for a foundation model on every AI task is a reflex that gets expensive fast, because a lot of what teams build is a well-defined single-purpose job that a managed AWS service already does cheaper, faster, and deterministically. Textract, Comprehend, Transcribe, Translate, Rekognition, Polly, Lex, Personalize, and Kendra each own a task shape, and the win is knowing when one of them beats a prompt. The interesting cases are the pipelines where a purpose-built service feeds a foundation model rather than competing with it.

Read article 27 Jul 2026 · 39 min read

Exam Room · Advanced Generative AI Developer

Proving Where AI Content Came From

A generative feature raises four different questions at once: can you prove an output came from your model, can a reader see what the service is and is not meant to do, can you stop bad behaviour as it happens, and can you show how the model was built and evaluated. Each one has a different AWS control, and reaching for the wrong one leaves the real question unanswered. The job is matching the responsibility to the artefact that actually discharges it.

Read article 27 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Where Humans Belong in a GenAI Pipeline

A generative-AI system needs human judgement in three different places, and it is easy to muddle them because all three sound like putting a person in the loop. Labelling training data, reviewing live predictions, and judging model quality are separate jobs with separate shapes, and with AWS's managed labelling and review services now in maintenance, two of them are loops a new build assembles itself. Getting the placement right keeps people on the decisions only people can make and keeps them off the ones a threshold or a metric already handles.

Read article 27 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Searching Images and Text With Multimodal Embeddings

Searching a product catalogue by typing a description, or by dropping in a lookalike photo, sounds like a job for a vision model, but a chat model that reads an image reasons about it rather than retrieving against a library. The tool that powers cross-modal search is a multimodal embedding model, which places images and text in one shared vector space so nearest-neighbour finds matches whichever modality the query arrived in. Getting it right means separating retrieval from reasoning, and keeping the distance metric matched to the model.

Read article 27 Jul 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

Tuning How a Model Samples: Temperature, Top-P, and Top-K

Temperature, top-p, and top-k all change how a model picks its next token, and reaching for the wrong one turns a crisp classifier into a coin flip or a brainstorm into a monotone. Each parameter shapes the distribution in a different place, they interact in ways that surprise people who turn all three dials at once, and the right setting is a property of the task rather than a house style. The job is knowing which knob does what and leaving the rest alone.

Read article 27 Jul 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Budgeting Tokens for a Long-Document Workload

A workload that reads long documents keeps hitting the context window and the bill keeps climbing, and the instinct is to reach for the model with the biggest window and stuff whole files into it. That instinct is expensive, slower, and can quietly lower answer quality as the relevant facts get buried. The real job is reasoning about tokens: what fits, what it costs on the way in and on the way out, and how to shrink the input without losing the part that matters.

Read article 27 Jul 2026 · 22 min read

Exam Room · Advanced Generative AI Developer

SageMaker JumpStart or Bedrock for the Same Model

A team that wants Llama 3.3 70B in production can take it through SageMaker JumpStart, pushing a button to deploy onto an endpoint they own, or through Bedrock's model catalog with per-token pricing and no infrastructure. Same model, sometimes the same base weights, two quite different operational shapes. The correct choice depends on how much of the serving layer you actually want to touch.

Read article 27 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Making an LLM Output Reproducible

Two people run the same prompt through the same model and get two different answers, and now a compliance reviewer wants to know which one the system actually produced. Temperature zero helps, pinning the model version helps more, and only a cache in front of the model gives you byte-identical repeats. The real question is how much determinism the use case genuinely needs before you pay for it.

Read article 28 Jul 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Why Your RAG Returns the Wrong Chunk

A retrieval system answers with confidence and cites the wrong passage, and the reasons are frustratingly varied: a distance metric that disagrees with the embedding model, chunks too big or too small, a query phrased nothing like the documents, an SKU that semantic search cannot match, a missing tenant filter, no reranking, or an index that never re-synced. This walks each symptom back to its cause and lands it on a fix.

Read article 28 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Metadata Filtering for Multi-Tenant Retrieval

One retrieval index serves every tenant, so the question that decides everything is whether a query can ever surface a document its asker was never meant to see. Metadata filters keyed on verified identity are the boundary; a prompt that asks the model to stay in its lane is not. This walks through attaching tenant and access metadata at ingestion, filtering at query time, and why pre-filtering beats filtering the results afterwards.

Read article 28 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Parent-Document Retrieval: Small Chunks, Big Context

The chunk size that retrieves precisely and the chunk size that gives the model enough context pull in opposite directions: small chunks embed cleanly and match the query but hand the model a fragment, while large chunks carry the surrounding context but blur the embedding and match worse. The fix is to stop treating them as one number and decouple the unit you search from the unit you return, which is what parent-document retrieval, hierarchical chunking, and sentence-window retrieval each do.

Read article 28 Jul 2026 · 32 min read

Exam Room · Advanced Generative AI Developer

Agentic RAG: When Retrieval Needs to Reason

A support assistant that answered one kind of question from one knowledge base now gets questions that need two lookups, a reformulated search, and a second retrieval once the first one comes back thin. Plain RAG runs a fixed pipeline: embed the query, fetch the top matches once, generate. Agentic RAG moves the retrieval decisions to the model, which can choose whether to search at all, which source to hit, how to rewrite the query, and whether to go back for more. The first is cheap and predictable; the second is the answer when one pass genuinely cannot get there, and a needless tax when it can.

Read article 28 Jul 2026 · 40 min read

Exam Room · Advanced Generative AI Developer

Designing Safe Tool Schemas for an AgentCore Gateway

A gateway turns your Lambda functions and REST APIs into tools a foundation model can call on its own judgement, and every tool you publish is a capability you are granting. The arguments arrive already shaped by a model that can be steered by whatever text reached the prompt, so a tool that takes a free-form instruction lets one coaxed call reach much further than you meant. How much the schema can rule out varies with how the tool is attached, the caller's identity does not travel with the call unless you arrange it, and the confirmation gate is code you own rather than a flag you set.

Read article 28 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Surviving a Model Deprecation on Bedrock

A foundation model you pinned two quarters ago will not live forever; Bedrock versions its models, marks the old ones legacy, and eventually retires them. Pin a version and behaviour stays stable, but the migration bill comes due on a schedule you do not control. Chase the newest version and you trade that bill for silent behaviour drift. The way through is a rehearsed migration: pin, watch the notices, test the successor against a saved eval set, and cut over behind a flag with a rollback ready.

Read article 28 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Measuring Hallucination in a RAG System

A retrieval-augmented assistant that answers confidently and wrong is failing in one of two very different places, and the fix depends on which. A hallucination can come from retrieval, where the context never held the answer, or from generation, where the model ran past what the context said, and a single quality number hides both. Measuring it well means separating faithfulness to the retrieved passages from whether the answer is even the right one, then attributing each miss to the stage that caused it.

Read article 29 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Choosing an Embedding Model for a Multilingual Corpus

A support knowledge base written in English, French, and Japanese needs semantic search, and the obvious move of running an English embedding model over everything quietly fails: a French question never finds the English answer that solves it. Multilingual embedding models put every language into one shared vector space so a query in one language retrieves documents in another, but only some models do it, and the choice turns on language coverage, whether cross-lingual matching is needed at all, embedding dimension, and how much text fits in one call.

Read article 29 Jul 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

LLM-as-a-Judge: Designing a Rubric You Can Trust

Handing evaluation to a second model is fast and it scales, but a judge with a vague rubric or an unchecked bias just launders its own preferences into a number you then trust. Pointwise scoring against explicit criteria and pairwise comparison answer different questions, and both leak position, verbosity, and self-preference bias unless you design against it. The work is writing a concrete rubric, controlling the known biases, and calibrating the judge against human labels before you let it grade at scale.

Read article 29 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Monitoring a Production Bedrock App

A generative-AI feature on Amazon Bedrock breaks in three directions: the bill climbs, the responses slow down, and the answers quietly get worse. The platform hands you metrics and logs for the first two almost for free, but quality is the one signal Bedrock cannot measure for you, so you have to build it. Knowing which of the three you are looking at, and where the signal lives, is the whole job.

Read article 29 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Building Deterministic Pipelines With Bedrock Flows

When a generative workflow has a sequence you already know, letting a model pick the order each time is the wrong trade: nondeterministic, harder to test, and an extra model call at every decision point. Amazon Bedrock Flows lets you draw the sequence as a fixed graph of nodes wired together with data links, so a designer owns the control flow and the model does the work inside a step. This walks the space between a Flow, a single agent, and Step Functions, and lands on when the drawn graph is the right home.

Read article 29 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Encrypting a Bedrock App End to End With KMS

Everything a Bedrock app stores is encrypted at rest by default, so the question is never whether the data is encrypted; it is who holds the key that decrypts it. A customer-managed KMS key turns encryption from a checkbox into a control you own: a lock you set the policy on, an audit trail of every use, and a switch you can throw to cut access without deleting a single byte. This walks the persistent artefacts one by one and asks, for each, whose key it is.

Read article 29 Jul 2026 · 33 min read

Exam Room · Advanced Generative AI Developer

Event-Driven GenAI: Processing Documents Asynchronously

Summarising a 200-page report or enriching a batch of records is a job that takes minutes, so wiring it behind a synchronous HTTP request guarantees timeouts, retries that re-run expensive work, and a bill that scales with impatience. The fix is to make the upload an event and let the work happen in the background. S3 notifications, SQS buffering, Lambda workers, Step Functions pipelines, and Bedrock batch inference each fit a different point on the volume-and-latency curve, and picking the wrong one shows up as either idle capacity or dropped documents.

Read article 29 Jul 2026 · 21 min read

Exam Room · Advanced Generative AI Developer

Caching LLM Responses Without Stale Answers

Thirty percent of the support assistant's queries are paraphrases of each other, 'how do I cancel?' 'can I cancel?' 'where's the cancel button?', and every one costs full model price. Caching LLM responses isn't as simple as hashing a prompt: exact-match, semantic, and prefix caching answer different questions, and getting the boundary wrong serves yesterday's answer to today's question.

Read article 29 Jul 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

A/B Testing Prompts and Models in Production

A new prompt scores better on the test set, a newer model is cheaper per token, and both look like obvious upgrades until they meet real traffic and quietly regress a slice of it. Offline scores tell you a variant is plausible; only live traffic tells you it is better. The job is deciding how much risk each change deserves, shadowing or splitting accordingly, measuring quality and latency and cost together, and rolling the winner out in a way you can undo in seconds.

Read article 30 Jul 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Right-Sizing Provisioned Throughput for a Custom Model

Some models, once you fine-tune them on Amazon Bedrock, cannot be called on demand at all; the only way to serve them is Provisioned Throughput, bought in model units you reserve and pay for whether traffic fills them or not. Size it from peak tokens per minute with headroom, then choose a commitment term, because a six-month lock is cheap per unit and hard to escape if demand moves. Over-provision and you burn money on idle units; under-provision and you throttle real requests. The sizing is the whole game.

Read article 30 Jul 2026 · 32 min read

Exam Room · Advanced Generative AI Developer

Defending Against Indirect Prompt Injection in RAG

A retrieval-augmented assistant answers from a knowledge base you built over your own documents, so the retrieved context feels trusted. It is not. The moment a document can be edited by a partner, crawled from the web, or fed in from user-generated content, an attacker can plant instructions that ride into the model alongside the genuine text, and the model cannot tell your data apart from a command hidden inside it.

Read article 30 Jul 2026 · 37 min read

Exam Room · Advanced Generative AI Developer

Cost Attribution and Tagging for GenAI Workloads

A single Bedrock bill arrives as one on-demand line, and the finance team wants to know which team, feature, and customer spent it. Application inference profiles, cost allocation tags, model invocation logging, Cost Explorer, and Budgets each pin the spend at a different grain, from whole account down to a single request. The job is choosing the grain that answers the chargeback question, then wiring the alert that closes the loop before the next bill lands.

Read article 30 Jul 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Building a Voice Assistant: Transcribe, Bedrock, and Polly

A voice assistant is really four pieces bolted together: speech becomes text, a model reasons over the text, the reply becomes speech, and something manages the back-and-forth. The trap is picking each piece in isolation and discovering at integration time that the latency has stacked up past what a caller will tolerate, or that a name has landed in a model prompt that a compliance rule says should never have been kept. Getting it right means budgeting latency across the whole chain and putting safety on the text stage, where the words actually live.

Read article 30 Jul 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

Handling Throttling and Rate Limits Gracefully

A Bedrock feature that ran clean in testing starts returning ThrottlingException the week traffic doubles, and the temptation is to bolt on a retry loop and move on. Retries smooth a transient spike, but they add no capacity, so a genuinely undersized workload just retries its way into a slower failure. The real work is telling a transient limit from a structural one, then reaching for backoff, a quota increase, cross-region inference, provisioned throughput, or a queue depending on which one the traffic actually needs.

Read article 30 Jul 2026 · 35 min read

Exam Room · Advanced Generative AI Developer

Cutting Ingestion Cost by Caching and Batching Embeddings

A retrieval system re-embeds its whole corpus on every update, and the bill scales with a knowledge base that barely changes. Embedding is a per-token cost paid on every chunk you send, so re-embedding text that has not changed is money spent to arrive at the vector you already had. Incremental sync, content-hash caching, de-duplication, request batching, and a smaller embedding dimension each cut a different slice of that waste, and each carries a bookkeeping cost worth naming before you reach for it.

Read article 30 Jul 2026 · 21 min read

Exam Room · Advanced Generative AI Developer

Choosing Between Kiro, Amazon Quick, and Bedrock

AWS sells finished AI assistants and it sells the platform to build your own, and most bad decisions in this corner come from confusing the two. Kiro is the assistant developers adopt; Amazon Quick is the assistant staff use over enterprise data; Bedrock is the platform a customer-facing feature gets built on. Sort by who the output is for and the choice mostly makes itself.

Read article 31 Jul 2026 · 32 min read

Exam Room · Advanced Generative AI Developer

Keeping a Knowledge Base Fresh Without Re-Embedding Everything

A retrieval assistant is only as good as the documents behind it, and those documents keep changing while the embedding bill for re-processing them keeps climbing. A support team on Amazon Bedrock has a Knowledge Base over forty thousand policy and product files, and their nightly full re-sync costs more than the queries it serves. The fix is knowing what actually needs re-embedding, when to trigger the work, and when retrieval is the wrong tool for the fact entirely.

Read article 31 Jul 2026 · 22 min read

Exam Room · Advanced Generative AI Developer

How to Wire Function Calling Through Bedrock

An assistant that knows the answers but can't act on them is half a tool. Function calling, letting the model invoke tools we define, with arguments it chooses, turns understanding into action. Bedrock's Converse API has native tool-use support; so does Anthropic's Messages API via Bedrock; so does the AgentCore harness, which runs the loop for you. Each exposes function calling through a different surface, and picking wrong makes the simple case hard.

Read article 31 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

When to Orchestrate With Step Functions Instead of an Agent

A multi-step GenAI job can be run three ways: a Bedrock agent that lets the model decide the sequence, a Bedrock Flow that draws the sequence as a fixed graph, or a Step Functions state machine that runs the sequence as a durable workflow with the model as one step among many. The deciding line is whether the control flow is known ahead of time and how much durability, retry, and cross-service reach the job needs. This walks the space and lands on when the state machine is the right home.

Read article 31 Jul 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Building a Golden Dataset for LLM Evaluation

Every prompt tweak, model swap, and RAG change gets judged against something, and if that something is a handful of screenshots in a chat thread, the judgement is noise. A golden dataset is the fixed yardstick: representative inputs paired with accepted answers, covering the easy middle, the known-hard cases, and the questions the system should refuse. Building one that stays trustworthy and reusable is the work that makes every later evaluation mean something.

Read article 31 Jul 2026 · 14 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Model Selection and Inference

A condensed revision sheet for choosing a model on Amazon Bedrock and picking how it runs: providers, the Nova family, selection axes, and every inference mode from on-demand to Provisioned Throughput to SageMaker hosting.

Read article 31 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Red-Teaming a Bedrock Application

A team ships a generative-AI assistant on Bedrock and waits to see what breaks. Red-teaming flips that around: you attack your own application on purpose, before an outsider does, probing for jailbreaks, injected instructions, leaked data, and tools that can be talked into doing damage. The work that lasts is not the single afternoon of creative attacks; it is turning every finding into a test that runs forever.

Read article 31 Jul 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Content Moderation With Rekognition, Comprehend, and Guardrails

A single generative app now takes uploaded photos, voice notes, and free text, and it also emits model output that has to be safe to show. One moderation service cannot cover all of that, because the right tool is decided by the media and by the stage of the pipeline. Rekognition reads images and video, Comprehend reads text, Transcribe bridges audio into text, and Bedrock Guardrails sits on the model prompt and completion. The job is routing each piece of content to the service built for it rather than forcing one tool to do everything.

Read article 31 Jul 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Cost Guardrails: Budgets, Quotas, and Model Choice

A generative-AI feature can double its own bill between two invoices, and the first anyone hears of it is the invoice. Per-token pricing, model tier, replayed context, call volume, and Provisioned Throughput commitments are the dials that set the number, and every one of them can be bounded in advance. The job is designing the spend limits in up front, budgets and quotas and model choice together, so an overrun trips an alert on day two rather than a surprise on the statement.

Read article 01 Aug 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Routing Requests Between a Cheap and a Capable Model

Most traffic to an LLM feature is easy: a label, a short extraction, a one-line answer that a small cheap model handles as well as the flagship. A minority needs the big model to reason through several steps. Paying flagship prices on every request treats the whole workload as if it were the hard tail, so the money and the latency go where they are not needed. Routing splits the stream, and the risk is a misroute that sends a hard request to the weak model and gets a confident wrong answer.

Read article 01 Aug 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Summarising Long Conversations to Fit the Context Window

A support copilot on Bedrock works fine for the first dozen turns, then the transcript it replays on every call grows until the bill climbs and the oldest messages fall off the edge of the context window. Replaying every turn is the simple thing, and it is the thing that breaks first. Sliding windows, running summaries, and extracted facts each trade history for room in a different way, and the split between a recent transcript and a durable store is the line between short-term and long-term memory.

Read article 01 Aug 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

Choosing an Embedding Dimension and Its Storage Cost

Embedding dimension looks like a knob where higher is simply better, but every extra dimension multiplies across every vector in the index, and storage plus search cost climb in lockstep. Amazon Titan Text Embeddings v2 lets you ask for 1024, 512, or 256 dimensions, so the choice is a real trade of retrieval quality against footprint and latency. The job is to size the corpus, set a quality bar, and measure what each dimension is actually worth on your own data.

Read article 01 Aug 2026 · 9 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Prompt Engineering

A condensed revision sheet for prompt engineering on Bedrock: techniques, decision rules, traps, and one-line facts for the AIP-C01 Generative AI Developer track.

Read article 01 Aug 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Tracing an Agent's Decisions in Production

An agent told a subscriber they were owed the wrong refund, and the final answer says nothing about why. To debug it you have to reconstruct the run: which tool it called, with what arguments, what came back, and how it reasoned from one step to the next. AgentCore emits metrics for that out of the box but not the spans, model invocation logging captures the prompts and completions, and distributed tracing ties the tool Lambdas into one picture. None of it is retroactive, which makes the setup a decision you make before the run you will need it for.

Read article 01 Aug 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Multi-Region Resilience for a GenAI Service

A Bedrock service is fine until the day its region has a bad hour, or until on-demand throttling caps it at the worst possible moment. Cross-region inference profiles, a warm second region with the same guardrails and knowledge base, replicated RAG data, and an honest read on which models the failover region actually supports all pull in the same direction. The decision comes down to the availability target, where the data is allowed to be processed, and what a second region costs to keep warm.

Read article 01 Aug 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Delivering Responses: Sync, Async, or Streaming

Every generative feature needs a way to get the answer to the user: one synchronous reply, a stream of tokens as they are generated, or an asynchronous job that finishes in the background. The wrong choice shows up as a spinner that stalls for thirty seconds, a gateway timeout on a long completion, or a batch of ten thousand documents jammed through a real-time endpoint one at a time. The deciding factors are how interactive the moment is, how long the output runs, and whether the caller can afford to wait.

Read article 02 Aug 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Preparing a Dataset for Fine-Tuning

Fine-tuning a foundation model lives or dies on the data you feed it, and most teams reach for volume when they should be reaching for consistency. A few hundred clean, representative, identically formatted prompt-completion pairs will out-teach a scraped heap of thousands. The work is shaping that set: the JSONL schema the model expects, a train and validation split with no leakage, PII and duplicates stripped, and a held-out set to judge the result. Get the data right and the job is almost a formality.

Read article 02 Aug 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Finding the Documents That Never Reached the Knowledge Base

A Knowledge Base sync reports success and the assistant still cannot answer a question whose answer is sitting in one of the source buckets. The job-level view says forty thousand documents scanned and reports no failure at all, because a document that was skipped, ignored, or failed to embed leaves no trace at that altitude. Ingestion observability is a separate feature from the invocation logging most teams have already turned on, and it is the one that names the file and the reason.

Read article 02 Aug 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Building Permission-Safe Retrieval on a Bedrock Knowledge Base

A retrieval-augmented system lives or dies on its retrieval layer, and for years the easy answer for a permission-sensitive corpus was Amazon Kendra: managed connectors, mature relevance, document-level access control out of the box. Kendra closed to new customers in July 2026, so a team starting today builds that layer on a Bedrock Knowledge Base and owns the permission enforcement itself, or buys the finished assistant instead. This is what the build actually involves, and when buying beats building it.

Read article 02 Aug 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Extracting Structured Data From Documents at Scale

A pile of invoices, forms, and scanned contracts has to become clean database rows, and the tool choice decides everything downstream. Amazon Textract reads layout and text deterministically; a foundation model on Bedrock reasons over meaning but needs validating; Bedrock Data Automation wraps the whole pipeline; and the reliable pattern for most real documents combines them. The job is matching the tool to whether the work is reading pixels or understanding meaning.

Read article 02 Aug 2026 · 13 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: RAG and Vector Stores

A condensed revision sheet for retrieval-augmented generation on AWS: how documents get chunked and embedded, where the vectors live, how retrieval and reranking are tuned, and which managed service does the plumbing for you.

Read article 02 Aug 2026 · 26 min read

Exam Room · Advanced Generative AI Developer

Choosing a Model for Code Generation

A team wants a code-generation feature: complete a function, explain a stack trace, translate a service between languages, review a diff. On AWS that splits into two very different projects, driving a general foundation model on Amazon Bedrock yourself, or adopting Kiro, the managed coding assistant. The pick turns on whether you are building a feature or buying an assistant, how much the model must know your own codebase, and how the generated code gets checked before anyone runs it.

Read article 02 Aug 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Building a Feedback Loop From Users to Model Improvement

A thumbs-down button is easy to add and easy to waste. The reactions users leave, explicit ratings and corrections plus the quieter signals of an edit, a retry, or an abandoned session, only matter if they flow somewhere that changes model behaviour. This scenario builds the loop from raw reaction to measured, validated improvement, and keeps the feedback data governed the whole way through.

Read article 02 Aug 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

Open-Weight or Proprietary: Choosing How You Host a Model

A managed proprietary model on Amazon Bedrock is running in an afternoon and bills by the token, while an open-weight model you host yourself hands you the weights, deep customisation, and a per-hour cost curve that only pays off once traffic is steady. The choice is really about control versus managed simplicity, and about which cost curve fits the load. Bedrock, SageMaker, and Custom Model Import stake out the ground in between, so the pick turns on how much of the stack the team actually wants to own.

Read article 03 Aug 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

How Many Chunks to Retrieve: Tuning Top-K

Top-k is the one RAG knob everybody leaves at its default, and the default is almost never right for a given corpus. Set it too low and the passage that holds the answer never reaches the model, so it hedges or invents. Set it too high and you pay for input tokens, add latency, and bury the good chunk among distractors the model reads right past. The number that works depends on chunk size, corpus, and question shape, and the only way to find it is to measure on your own eval set.

Read article 03 Aug 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Grounding on Fresh Data: Tools or RAG

A model that answers from its own weights is answering from a frozen snapshot, so anything that has changed since training is a guess. Two grounding patterns fix that in different ways: retrieval pulls passages from an indexed corpus, and a live tool call fetches the exact current value at request time. They solve different problems, and the wrong one gives you a confident answer built on stale data. The deciding questions are how fast the data moves, whether the answer is a passage or a precise number, and whether it is shared or per-user.

Read article 03 Aug 2026 · 35 min read

Exam Room · Advanced Generative AI Developer

Designing a Bot-to-Human Escalation Path

A generative assistant that never hands off will one day answer, with total confidence, the single question it should have escalated, and that is where the real damage lands. The signals that a bot has hit its limit are knowable: low confidence, a guardrail intervention, an out-of-scope or high-stakes request, a frustrated customer, or an action beyond its authority. Each of those maps to a clean route out. The design job is deciding what the bot must never settle alone, then wiring the handoff so a human inherits the full context instead of a cold start.

Read article 03 Aug 2026 · 12 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Agents and Orchestration

A one-page revision sheet for agents, tool use, and orchestration on Bedrock: when the model decides the flow, when you do, and which service owns each shape of workflow.

Read article 03 Aug 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Reducing End-to-End Latency in a GenAI App

A generative-AI feature that feels slow is rarely slow in one place; the seconds are spread across retrieval, the model, tool calls, and the network, and teams routinely tune the part that is easiest to see rather than the part that costs the most. The move that works is to measure each stage as p50 and p99, find the biggest contributor, and attack that one with the lever it responds to. Streaming, prompt caching, a smaller model, and parallel retrieval each help a different stage, so picking the wrong lever spends effort where the clock is not.

Read article 03 Aug 2026 · 33 min read

Exam Room · Advanced Generative AI Developer

Governing Model Access Across Many Teams

Fifty engineers across a dozen teams all want to build on Amazon Bedrock, and the platform team has to decide who can use which models, keep the whole organisation inside a shared policy, and still tell finance what each team actually spent. Enabling a model, scoping an IAM role, writing a service control policy, and tagging an inference profile are four different levers. The trick is to pull them so access stays least privilege at the granularity of a single model while cost and policy stay visible from the centre.

Read article 03 Aug 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Versioning and Rolling Back Prompts and Models

A team ships a prompt tweak on Friday and by Monday quality has quietly dropped, but nobody can say what changed because the prompt, the model reference, and the guardrail were all edited in place. The fix is to treat every piece of a generative feature as a versioned release artefact: a pinned model version, a published prompt version, a published guardrail version, an agent alias, all rolled out together and reversible together. Then a change is deliberate, a rollback is one repointed alias, and you always know exactly what moved.

Read article 04 Aug 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Choosing a Vector Index: HNSW, IVF, and the Trade-Offs

A retrieval feature slows to a crawl as the corpus grows, and the fix is almost never a bigger box; it is the index. Exact search gives perfect recall and dies at scale, HNSW gives fast high-recall queries for a lot of memory, and IVF trades a little recall for cheaper builds and a smaller footprint. The parameters that tune each one, m and ef_search and nprobe, all pull on the same three-way rope between recall, latency, and cost, and the only honest way to set them is to measure on your own data.

Read article 04 Aug 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

Dense, Sparse, or Hybrid Retrieval

Dense embeddings capture what a query means; sparse keyword search catches the exact token a query names. Pick the wrong one and a RAG system either cannot find a paraphrase or cannot find a part number, and which failure you hit depends entirely on the shape of your corpus and the shape of your queries. Hybrid runs both and fuses the scores, which is why it is the safe default for a mixed corpus, but it costs two retrievers and a fusion step that has to be tuned.

Read article 04 Aug 2026 · 35 min read

Exam Room · Advanced Generative AI Developer

Handling Ambiguous Questions With Clarification

A user asks a vague question and the model answers it as if it were perfectly clear, confidently and often wrongly. The fix is not a bigger model; it is teaching the assistant to notice when it lacks enough to answer, then decide between asking a clarifying question, offering the likely readings, or quietly resolving the gap from account data and prior turns. Over-asking annoys, over-assuming misleads, and the balance depends on how costly a wrong answer is.

Read article 04 Aug 2026 · 11 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Model Customisation

A revision sheet for customising a foundation model on Bedrock and SageMaker: the escalation ladder from prompting to fine-tuning to continued pre-training, dataset prep, hyperparameters, and how the custom model actually gets served.

Read article 04 Aug 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Preventing Data Exfiltration Through an LLM

A generative-AI assistant sits on top of an internal knowledge base and a set of tools, and every one of those connections is a way for data to leave. Retrieval can surface a document the user should never see; a tool can return more rows than the user is entitled to; the model can repeat a secret someone put in its context. The uncomfortable part is that you cannot fix any of it by telling the model to keep quiet, because access control belongs in retrieval and tools, not in the prompt.

Read article 04 Aug 2026 · 30 min read

Exam Room · Advanced Generative AI Developer

Deciding Whether to Use GenAI at All

A generative model is the most flexible tool in the box and the most expensive, the least predictable, and the hardest to validate. Plenty of tasks that get handed to an LLM are better served by a rule, a lookup, a classic classifier, or a purpose-built AWS service that is cheaper, faster, and deterministic. The job is matching the tool to the task rather than defaulting to a model because it is the exciting one.

Read article 04 Aug 2026 · 33 min read

Exam Room · Advanced Generative AI Developer

Choosing Where to Store Conversation State

A chat assistant makes stateless model calls, so the transcript and session data have to live somewhere between turns, and the somewhere decides latency, cost, and how much of the assistant you end up building yourself. DynamoDB, ElastiCache, MemoryDB, and managed agent memory each fit a different mix of durability, turn rate, and how much you want to own. The job is matching the store to the conversation shape rather than reaching for the one you used last time.

Read article 05 Aug 2026 · 25 min read

Exam Room · Advanced Generative AI Developer

Choosing a Distance Metric for Embeddings

A retrieval system can look completely healthy, index built, queries fast, no errors in the logs, and still quietly return the wrong neighbours because the distance metric does not match what the embedding model produces. Cosine, dot product, and Euclidean each measure closeness differently, and the vector store makes you pick one. The job is matching that choice to the model that generated the vectors, because the wrong metric degrades recall without ever raising an alarm.

Read article 05 Aug 2026 · 35 min read

Exam Room · Advanced Generative AI Developer

Cutting Cost per Query in a RAG System

A retrieval-augmented query looks cheap until the monthly bill arrives, and most of the money is hiding in a place nobody profiles: the retrieved context, which rides along as input tokens on every single call. Cutting cost per query is less about a cheaper model and more about returning fewer and better chunks, caching what repeats, and paying the vector store only for what it holds. This walks the money through one RAG query and trims each part in turn.

Read article 05 Aug 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Writing a System Prompt for a Production Assistant

A production assistant needs standing instructions that hold steady while every request changes underneath them, and the system prompt is where they live. It can set the role, the tone, the refusals, and the output shape, and it can tell the model how to lean on retrieved context. What it cannot do is keep a determined user from talking the model out of all of it, so the design question is which jobs the system prompt should own and which need Guardrails, least-privilege tools, and delimiters standing behind it.

Read article 05 Aug 2026 · 14 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Evaluation, Cost, and Operations

A condensed revision sheet for running a generative AI app on AWS: how to evaluate quality, watch it in production, keep the bill sane, hold latency down, and ship without breaking things.

Read article 05 Aug 2026 · 29 min read

Exam Room · Advanced Generative AI Developer

Choosing a Guardrail Strategy: Managed, Custom, or Both

A generative-AI app needs safety controls, and the instinct is to reach for one tool and call it done. Amazon Bedrock Guardrails covers a lot out of the box: denied topics, content filters, PII redaction, grounding checks, prompt-attack detection. But some of the rules a business actually cares about are bespoke, and no managed policy covers them. The real design decision is which risks belong to the managed layer, which need custom code, and how the two stack.

Read article 05 Aug 2026 · 38 min read

Exam Room · Advanced Generative AI Developer

Taking a GenAI Feature From Proof of Concept to Production

A demo that works in a notebook is not a feature that survives production. The gap is a set of dimensions a proof of concept gets to skip: evaluation, safety, security, reliability, cost, observability, governance, and operations. A demo proves the feature is feasible; production proves it is safe, affordable, reliable, and measurable, and closing that gap is a checklist, not a rewrite.

Read article 05 Aug 2026 · 27 min read

Exam Room · Advanced Generative AI Developer

Chunking Code, Tables, and Mixed Content

Fixed-size chunking is fine for prose and quietly ruinous for everything else. It splits a function halfway through its body, tears a table away from its header row, and floats a figure loose from the caption that explains it, and every one of those breaks lowers the odds the right passage ever comes back from retrieval. The fix is to cut on the structure the content already has rather than on a token count that ignores it.

Read article 06 Aug 2026 · 39 min read

Exam Room · Advanced Generative AI Developer

Generating and Understanding Images, Audio, and Video on Bedrock

Non-text work on Bedrock splits cleanly into two jobs that people keep conflating: making pixels and audio with a foundation model, and pulling structure out of media somebody else made. Generation now runs through the Stability image models and Luma Ray 2, with Nova Canvas and Nova Reel marked Legacy in the catalogue; understanding runs through Bedrock Data Automation, a purpose-built service, or a multimodal model reasoning in a prompt. Pick the wrong half and you fight the platform the whole way. This walks the split, then routes by modality and by whether you need structured fields or free-form reasoning out the other end.

Read article 06 Aug 2026 · 16 min read

Exam Room · Advanced Generative AI Developer

Cheat Sheet: Security and Responsible AI

A final-review sheet for securing and governing generative AI on AWS: identity, network, encryption, data boundary, Bedrock Guardrails, injection defence, and the responsible-AI dimensions, condensed to what you need on the morning.

Read article 15 Aug 2026 · 28 min read

Exam Room · Advanced Generative AI Developer

How to Pay for Serving a Model on Bedrock

Three models are heading for production: a hosted foundation model, a fine-tune halfway through training, and a set of open weights from a research spike. Finance wants one forecast. Each bills on a different unit, and for two of them the unit was settled the moment somebody picked a base model. Serving cost is less a lever you pull afterwards than a constraint you inherit from where the weights came from.

Read article 15 Aug 2026 · 31 min read

Exam Room · Advanced Generative AI Developer

Choosing an Agent Framework for the AgentCore Runtime

Deciding to own the reasoning loop settles where the agent runs and leaves open what you write it in. AgentCore takes any framework, so the choice is not about compatibility; it is about how much of the runtime you get for free, whether the traces arrive without you building them, and what happens when one agent becomes four. Frameworks that already speak OpenTelemetry and the GenAI conventions hand you observability as a dependency rather than a project, and that turns out to separate the field more than the loop syntax does.

Read article 16 Aug 2026 · 34 min read

Exam Room · Advanced Generative AI Developer

Giving an Agent Credentials Without a Standing Key

An agent that reaches real systems needs credentials, and the quickest way to give it some is the one that ages worst: a long-lived key in the environment that every request shares. The alternative splits into two questions that get conflated. Establishing who is asking is inbound authorisation and happens before your code runs; obtaining something to call a downstream service with is outbound, and which flow you can use depends on whether the call is machine-to-machine, delegated by a user, or exchanged from a token you were already handed.

Read article 17 Aug 2026 · 35 min read

Exam Room · Advanced Generative AI Developer

A Domain-by-Domain Checklist for the Generative AI Developer Exam

The Generative AI Developer certification scores five domains, weighted from 31% down to 11%. This is the whole GenAI track on one page, sorted into those five domains: the cheat sheet that anchors each one, the decisions to work through, the pop quizzes to drill, and the labs to build. Tick your way down each domain, and wherever a line still feels shaky, that is where the next hour of revision goes.

Read article

13 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: OpenSearch Serverless's Hidden Floor

You need a fully-managed vector store for a Bedrock Knowledge Base. What is the hidden cost of OpenSearch Serverless?

Read article 14 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: When Aurora pgvector Wins

When is Aurora PostgreSQL with pgvector the better vector store for RAG?

Read article 15 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: The Silent Distance-Metric Bug

Retrieval quality is poor even though the embeddings look fine. What silent config is worth checking?

Read article 16 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: S3 Vectors and the Latency Budget

You have a huge, rarely-queried vector archive where seconds of latency is acceptable. Cheapest fit?

Read article 17 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Why Hybrid Search Finds ERR-4021

Semantic search keeps missing exact tokens like an error code. What retrieval change fixes it?

Read article 18 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: What a Reranker Actually Fixes

The right chunk is retrieved but ranks twelfth, below the cutoff. What promotes it?

Read article 19 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Hierarchical Chunking in One Line

Answers get cut across chunk boundaries in a long structured PDF. Best Bedrock KB chunking?

Read article 20 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Freshness and Access Are Metadata

Retrieval must respect access control and document freshness. What lever, cheaply?

Read article 21 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: The Prompt-and-Completion Record

An auditor asks to see every prompt this assistant received last Tuesday. What produces it?

Read article 22 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Who Changed It vs What It Said

Who turned off the PII filter, and when? Which log, and why not invocation logging?

Read article 23 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Turning Logs Into an Audit

The reviewer wants a control-mapped report and a statement of what the model is approved for. Two artifacts?

Read article 24 Jul 2026 · 5 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: The Eight Responsible-AI Dimensions

AWS names its responsible-AI dimensions. Roughly, what are they?

Read article 25 Jul 2026 · 4 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Packing Many Models Onto One Endpoint

How do you serve dozens or hundreds of models without an endpoint each?

Read article 25 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Measuring Bias With fmeval

You must measure a GenAI feature for bias and toxicity before launch. Which tool?

Read article 26 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: What Guardrails Enforce

What does Bedrock Guardrails actually enforce at runtime?

Read article 27 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: LLM Explainability Is Traceability

Someone asks you to explain a RAG answer. What is the realistic form of FM explainability?

Read article 28 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Faithful but Wrong

A RAG answer scores high on faithfulness but is still wrong. How?

Read article 29 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Evaluating Both Halves of RAG

You cannot tell if a wrong RAG answer is a retrieval or a generation problem. What evaluates each half?

Read article 30 Jul 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: LLM-as-a-Judge, and the Catch

You need to score thousands of outputs on quality without a human reading each. Approach and caveat?

Read article 31 Jul 2026 · 4 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Letting an LLM Take Actions

Your LLM needs to take real actions, like calling an API. What wires that on Bedrock?

Read article 01 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Turning Down the Randomness

Outputs are too random for a structured extraction task. Which inference parameters, and which way?

Read article 02 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Provisioned Throughput vs On-Demand

Steady high-volume Bedrock traffic with latency guarantees. Provisioned Throughput or on-demand?

Read article 03 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Prompts Are Versioned Assets

Thirty services share prompts and you need versioning and reuse. What on Bedrock?

Read article 04 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Cutting the Bill Without Losing Quality

Your Bedrock bill is high but quality must hold. First levers?

Read article 05 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Seeing Into a Production Bedrock App

What gives you operational visibility into a production Bedrock app?

Read article 06 Aug 2026 · 3 min read

Exam Room · Advanced Generative AI Developer

Pop Quiz: Keeping PII Out of Prompts and Logs

Keep customer PII out of prompts and logs. What is the built-in control?

Read article

27 Jul 2026 · 13 min read

Exam Room · Advanced Generative AI Developer

Lab: Invoke a Foundation Model From Lambda

A hands-on lab, the first in the series. You are handed a Lambda function with an execution role, a model id in an environment variable, and one gap: the Bedrock call itself. Five lines of Python, deployed with CloudFormation, proven with a script, torn down when you are done. Every later lab builds on this function, so this is where the track starts.

Read article 28 Jul 2026 · 11 min read

Exam Room · Advanced Generative AI Developer

Lab: Put a Guardrail in Front of a Bedrock Model

A hands-on lab. You are handed a working Bedrock model call and a fully-configured guardrail, and your job is the two lines that connect them. Deploy real infrastructure with CloudFormation, block a financial-advice question, watch a phone number get redacted, then tear it all down. The reading tells you what a guardrail is; this makes you wire one in.

Read article 29 Jul 2026 · 11 min read

Exam Room · Advanced Generative AI Developer

Lab: Get Structured JSON Out With Tool Use

A hands-on lab. Asking a model for JSON in the prompt and hoping is how parsers break at 3am. This lab has you do it the reliable way: declare a tool schema, let the model fill it in, and read the answer back as parsed arguments rather than a string. You deploy a Lambda, turn a free-text support message into a structured record, and find out how much of the shape the schema really guarantees.

Read article 30 Jul 2026 · 10 min read

Exam Room · Advanced Generative AI Developer

Lab: Give a Bedrock Chatbot a Memory

A hands-on lab. A model call is stateless: ask a chatbot your name, ask it back, and it has already forgotten. Memory is not a feature you switch on, it is a transcript you replay, stored somewhere between requests. This lab hands you a Lambda and a DynamoDB table and has you build the load-and-save loop that turns single calls into a conversation, then reason about what replaying all those turns costs.

Read article 31 Jul 2026 · 8 min read

Exam Room · Advanced Generative AI Developer

Lab: Build RAG From Scratch

A hands-on lab. A Knowledge Base hides retrieval behind an API; this lab makes you build the thing it hides. Five documents, a model, and one function you write: embed the question, compare it to the documents, ground the answer in the closest ones. No vector store, so the mechanics of RAG are in plain sight, and the concept stops being magic.

Read article 01 Aug 2026 · 9 min read

Exam Room · Advanced Generative AI Developer

Lab: Wire a Tool the Model Can Call

A hands-on lab. In an earlier lab the model filled in a schema and stopped; here it goes further, deciding on its own to call a tool, waiting for the result, and answering from it. That decide-call-read-continue cycle is exactly what a managed agent runs under the hood. You build it by hand, so an agent's tool wiring stops being a black box.

Read article 02 Aug 2026 · 9 min read

Exam Room · Advanced Generative AI Developer

Lab: Build a Data-Quality Gate

A hands-on lab. Bad data is the cheapest thing to stop at the door and the most expensive to fix after it is embedded. This lab hands you a pipeline that reads raw records from S3, routes them, and reports; you write the rules that separate the records fit to feed a model from the ones that would poison it. It is the hand-built version of a managed data-quality gate.

Read article 03 Aug 2026 · 10 min read

Exam Room · Advanced Generative AI Developer

Lab: Answer a Metric Question With Text-to-SQL

A hands-on lab. You cannot embed your way to a SUM. When the answer is a computed number, retrieval is the wrong tool and text-to-SQL is the right one: hand the model a schema, let it write the query, run it safely, summarise the result. You deploy a Lambda with a small database baked in and write the one step that turns a question into SQL, behind a guard that refuses anything but a read-only SELECT.

Read article 04 Aug 2026 · 21 min read

Exam Room · Advanced Generative AI Developer

Lab: Generate the Weekly Box Art

A hands-on lab. Greenbox's marketing changes every week because the box does, and the weekly change already exists as data: the box manifest. This wires Stability AI Stable Image Core and Luma Ray 2 onto the end of that pipeline, so one JSON file becomes the box hero, the recipe card art, a short clip of each featured farm's produce, and a looping clip of how to prep the vegetable nobody recognises. You write the two calls; the honesty policy is already in the manifest.

Read article 04 Aug 2026 · 9 min read

Exam Room · Advanced Generative AI Developer

Lab: Evaluate the Pipeline

A hands-on lab. Every feature so far left the same question hanging: is it any good, and would a change help or hurt? You cannot tell from one reply. This lab builds the loop that answers it: a golden set, an automatic judge, and a score, so a prompt edit or a model swap becomes a number you can compare. You write the judge, the LLM-as-a-judge that turns a verdict into that number.

Read article 05 Aug 2026 · 20 min read

Exam Room · Advanced Generative AI Developer

Lab: Fine-Tune a Model and Read the Loss Curves

A hands-on lab. The theory gave you epochs, learning rate, and the shape of a healthy loss curve. This runs an actual Bedrock customisation job over a few hundred support replies, then pulls the two metrics files back out of S3 and asks you to read them. You write the part that turns those CSVs into a verdict: underfitting, healthy, or overfitting, and the step the run should have stopped at. There is a free path, because a training job costs real money and takes hours.

Read article 05 Aug 2026 · 7 min read

Exam Room · Advanced Generative AI Developer

Lab: The Capstone

A hands-on lab, and the last one. No gap to fill and no scaffolding: just data, a requirement, and an acceptance test. You assemble a grounded, guarded, honest assistant from the pieces you have built across the track, and prove it passes. This is the shape of a production GenAI feature, where the model is one component and everything around it is what makes it trustworthy.

Read article 06 Aug 2026 · 18 min read

Exam Room · Advanced Generative AI Developer

Lab: Stand Up a Bedrock Knowledge Base

A hands-on lab. The from-scratch lab made you build retrieval by hand; this one hands the same documents to a managed Knowledge Base and asks you to write the two calls that query it. Chunking, an S3 Vectors index, an ingestion job, and the difference between asking for chunks and asking for an answer.

Read article