Course overview

AI Engineering: LLMs, RAG & Agents

0 / 26 done0%

1. LLM Foundations

How LLMs actually work (the honest mental model)
Tokens & embeddings: cost, limits, and meaning-as-math
Transformers & attention, without the calculus
Temperature, top-p & the API knobs that matter
Choosing a model: the axes and the algorithm

2. Prompt Engineering That Works

The 5-part prompt skeleton & iteration loop
Few-shot prompting: show, don't tell
Chain-of-thought & reasoning models
Structured output: JSON you can actually parse

3. Building with the API

Chat API anatomy: roles, state & stop reasons
Streaming: the #1 perceived-speed lever
Tool calling: how LLMs take action
Cost & latency engineering

4. RAG: Retrieval-Augmented Generation

Why RAG — and why not fine-tune or long context
Chunking: the stage that silently makes or breaks RAG
Embeddings & vector databases in production
The retrieval pipeline & the grounded prompt
Hybrid search & reranking: the biggest quality jump
Evaluating RAG: retrieval vs generation

5. AI Agents

What agents are (and when NOT to build one)
The agent loop from scratch — 40 lines
Patterns: planning, reflection, verifiers, multi-agent
Guardrails: budgets, approvals, sandboxes, injection

6. Evaluation, Observability & Capstone

Evaluation: the skill that separates ship from demo
Production observability & safety
Capstone: a grounded agent, end to end

0% complete

AI Engineering: LLMs, RAG & Agents

Chat API anatomy: roles, state & stop reasons

This lesson is part of the full course

Enroll in AI Engineering: LLMs, RAG & Agents for lifetime access to every lesson, diagram and code template.

Enroll — ₹399 Read a free preview lesson instead

Lifetime access Instant

Previous Next lesson