// course · free · self-paced

AI Engineer Foundations

RAG, agents, evals and serving, taught by building each one in Python and then answering the questions interviewers ask about it.

by the endYou can build, measure and explain a retrieval pipeline, a tool-using agent, an evaluation loop and a cost-aware serving design, and you can answer the interview questions on each without memorising.

Sukhveer Kaur

Software Developer & AI Engineer

6

Modules

48 +5

Lessons (+coming)

~5 h

Total read

Intermediate

Level

Before you start: You write Python at work and have called an HTTP API. Module 0 covers the rest.

48 lessons

Start → Python Environment Setup for AI Agents: …

Module 0 — Setup and the Python you need
8 · ~35 min · optional
Optional. Skip if you already run virtual environments, type hints and async without thinking.
you canA working environment, API keys in place, and enough Pydantic and async to read every code sample in the course.
Module 1 — Retrieval-augmented generation
9 · ~58 min
Build a RAG pipeline from a list of strings to a vector database, and measure why most of them answer badly.
you canYou can explain chunking, embeddings, retrieval and re-ranking with a hit-rate number you measured yourself.
Module 2 — AI agents
14 · ~94 min
Write the agent loop by hand, then rebuild it on LangGraph with tools, memory, guardrails and an API.
you canYou can draw the reason-act-observe loop, explain where it fails (loops, bad tool calls, cost) and show an agent you deployed.
Module 3 — Evaluation and observability
5 · ~34 min
Stop guessing whether a change helped. Golden sets, deterministic checks, LLM judges, and a CI gate that blocks bad PRs.
you canYou can name the metrics (faithfulness, context precision and recall, pass rate), build a golden set, and gate a deploy on it.
Module 4 — Serving and system design
10 · ~67 min
Cost and latency move together. Caching, rate limits, scaling and a gateway, applied to an LLM feature, then deployed.
you canYou can design a model-serving path with caching, routing and an eval hook, name the lever that moves each number, and deploy it.
Module 5 — Put it on your CV
2 · ~10 min
Turn the four capstones into a portfolio and a prep plan for the interview rounds.
you canA portfolio of four working projects you can talk about, and a schedule for the five interview rounds.
1. AI Engineering Portfolio Projects That Get Interviewscoming soonLesson
2. AI Engineer Behavioral Interview: Presenting the Project You Builtcoming soonInterview
3. The AI Engineer Interview Prep Plan (2–4 Weeks)coming soonLesson
4. AI Engineer Interview Questions (2026): RAG & AgentsInterview · 5 min
5. AI Engineer Salary 2026: Real Numbers by LevelOptional · 5 min

Preparing for interviews specifically? The AI Engineer Interview Questions (2026): RAG & Agents hub lists every round; each module above ends with the questions for its round.

Progress is stored only in this browser; use "Continue on another device" to carry it over. Lessons marked "coming soon" are planned and not yet published.

AI Engineer Foundations

Module 0 — Setup and the Python you need

Module 1 — Retrieval-augmented generation

Module 2 — AI agents

Module 3 — Evaluation and observability

Module 4 — Serving and system design

Module 5 — Put it on your CV