Applications of Foundation Models

Designing an application around a foundation model (FM) starts with a shortlist of requirements that any candidate model must pass: what data it handles, which languages it speaks, how much it can read and write in one request, whether it can be customized, and how its cost and speed fit the…

Inference parameters are settings you send with each request to shape how a foundation model writes its answer, without retraining or changing the model. Some control how predictable or varied the wording is; others control how long the answer can run and where it stops. Exam scenarios describe a…

Retrieval Augmented Generation (RAG) gives a foundation model access to your own information at the moment it answers. Instead of retraining the model, the application looks up relevant passages from your documents and adds them to the prompt, so the answer is based on current, company-specific…

A vector database stores embeddings, the lists of numbers that capture what a piece of text or an image means, and finds the stored items closest in meaning to a query. Every RAG application needs one. AWS offers several services that can play this role, and the exam expects you to pick the one…

There are several ways to make a foundation model better at your task, and they differ enormously in cost, effort and time. From cheapest to most expensive, they are: in-context learning (putting instructions and examples in the prompt), Retrieval Augmented Generation (RAG), fine-tuning, model…

An AI agent is software that uses a foundation model to work toward a goal: it plans the steps, calls tools and systems to act, checks the results, and keeps going until the task is done. A single prompt produces one answer; an agent can carry out a whole process, such as handling a claim from…

A prompt is the input you send a foundation model to get a response, and most prompts are built from a small set of parts: an instruction, context, input data, a description of the output you want and, for image models, a negative prompt. This lesson takes a prompt apart, so you can tell which…

Prompting techniques are repeatable ways of structuring a prompt to get better answers without changing the model. This lesson covers the ones the exam names: zero-shot, single-shot and few-shot prompting, chain-of-thought prompting and prompt templates. Exam questions usually describe a task and…

Good prompts are rarely written in one go. They come from clear, specific wording, careful testing and steady refinement, with guardrails around the model for the cases a prompt alone can't handle. This lesson covers the habits behind reliable prompts and the benefits they bring: better answers…

Prompts can be turned against an application. Attackers write inputs that override a model's instructions, coax out harmful content, reveal hidden prompts or data, or plant malicious text where the model will later read it. This lesson names each risk the exam lists (hijacking, jailbreaking…

Once a team has a prompt that works, it needs to store it, reuse it, improve it safely and know which version is running in production. Amazon Bedrock Prompt Management does that: it stores reusable prompts with variables, lets you test variants side by side, and saves numbered versions you can…

A foundation model can be trained in four main ways: pre-training, fine-tuning, continued pre-training and distillation. Each one changes something different about the model and needs a different kind of data. For the exam, the skill is reading a scenario and naming the method that matches what…

Fine-tuning isn't one technique but a family of them. Each one changes a pre-trained model in a different way, and each needs a different kind of data. This lesson covers instruction tuning, domain adaptation, transfer learning and continued pre-training, and shows how to tell from a scenario…

A fine-tuned model can only be as good as the data it learns from. Preparing that data means curating it, labeling it, making sure it represents the real job, getting the size right, and governing who can see and use it. This lesson covers each step, plus reinforcement learning from human feedback…

Before a company trusts a foundation model with real work, it needs evidence that the model does that work well. There are three broad ways to get that evidence: score the model automatically against test data, ask people to rate its answers, or use a second model as a judge. Amazon Bedrock offers…

When a foundation model writes text, you need a number that says how good that text is. The exam names four ways to get one: ROUGE, BLEU, BERTScore and LLM-as-a-judge. The first three compare the model's output with a reference answer written by a person. The fourth asks another model to grade the…

Most real applications don't call a foundation model once and show the answer. A RAG system searches your documents first, an agent may call several tools, and a workflow chains many steps. When the final answer is wrong, you need to know which step failed. So evaluating an FM-based application…

A foundation model can score well on every quality metric and still fail the business. The summaries might be excellent, yet nobody saves any time; the chatbot might be helpful, yet customers still phone support. Business alignment means checking that an AI application moves the outcome it was…