AI Engineering Fundamentals.
The on-ramp to AI engineering. We rebuild modern LLM systems from first principles; tokens, prompts, RAG, agents, evaluation, and production; concept first, code second, no framework magic. The natural prerequisite to AI Engineering Professional.
Modules
10
Hours of teaching
~23
Hands-on labs
50
Portfolio projects
9
What you'll learn
The bricks every AI engineer needs first.
See LLM systems as a small set of bricks; not a pile of frameworks.
Build prompts, RAG, agents, and structured extraction without LangChain magic.
Defend your apps against prompt injection and jailbreaks from day one.
Evaluate LLM systems with the same rigor as classical software.
Ship a production-grade LLM API; a castle you fully own and can defend.
Module-by-module
Ten modules. From your first API call to a production LLM service.
Each module is a wall of bricks. Each brick is a single concept with its own focused mini-lab. Each module ends with a small castle you can run and inspect.
01Setup & Working Style for LLM Apps+
Project layout, environment, API keys, cost control, prompts-as-code; the working discipline every LLM app needs from day one.
02Understanding Large Language Models+
What LLMs actually are, how they're trained, what they can and can't do; the mental model behind every system you'll build.
03Tokenization & LLM Generation+
BPE tokenization, vocabularies, sampling, temperature, top-k, top-p, log-probs; how generation actually works under the hood.
04Prompt Engineering & Security+
System prompts, few-shot, chain-of-thought, structured outputs, jailbreaks, prompt-injection defenses; prompting that survives in production.
05Embeddings & Document Processing+
Embedding geometry, similarity, document loaders, chunking strategies; the data layer behind every retrieval system.
06Building RAG Systems+
Vector stores, retrieval pipelines, grounding, citations; RAG built brick by brick, without LangChain magic.
07AI Agents+
Tool use, function calling, planner-executor loops, memory, guardrails; what makes an agent actually work, and the failure modes nobody warns you about.
08Structured Extraction+
Schemas, function calling for extraction, validation, recovery; turning unstructured text into clean, typed data.
09LLM Evaluation+
Eval sets, LLM-as-judge, regression testing; the eval-driven development loop your prompts and pipelines need.
10Production Deployment+
FastAPI services, streaming, caching, cost control, observability; shipping an LLM app on a real budget.
Projects you'll add to your portfolio
Nine castles you can defend in an interview.
Real systems you rebuild from scratch, not notebook tours. Each one ships as a portfolio piece you own.
- 01
Mini-Transformer (from scratch)
- 02
Interactive LLM Playground
- 03
Intelligent Document Data Extractor
- 04
Multi-Step Reasoning Engine
- 05
Injection-Resistant Secure Chatbot
- 06
Chat-with-Docs RAG Application
- 07
Tool-Using AI Agent (no frameworks)
- 08
Automated Evaluation Pipeline
- 09
Production-Grade LLM API with FastAPI
37 mini-labs + 13 project labs = 50 hands-on exercises.
Built for
- • Software engineers brand-new to AI / ML / DL
- • Career switchers who want a serious starting point
- • Students filling gaps between school and industry
- • Anyone who wants to start AI engineering the right way
Not built for
- • Engineers already shipping LLM apps (skip to Professional)
- • People hoping to skip the math entirely
- • Folks looking for a no-code tutorial tour
- • “AI in 7 days” shortcut seekers
Enroll today
Start learning on Udemy.
Full access to all modules, labs, and portfolio projects. Enroll on Udemy. Or join the Brick Letter for updates on the rest of the track.
Free. Unsubscribe anytime. We don't do scarcity tricks.