AI Fundamentals for Developers¶
A crash course that demystifies modern AI for technical developers. You will not train models. You will understand every piece of jargon your AI-engineer colleagues use, and build a working RAG pipeline and agent loop by hand.
Who this is for: Senior developers, tech leads, and consultants of any stack who work with AI systems but don't build them from scratch.
Time: ~5 hours of instruction (six modules) plus ~4–5 hours of optional hands-on labs (eight labs + a capstone). Each module and lab carries its own time estimate so you can self-pace. Do the reading in one or two sittings; treat the labs as take-home practice.
Choose your path¶
Approach A — Concept-First (Bottom-Up)¶
Start from the science, build up to applications. Best if you want to understand how it works before how to use it.
Approach B — Use-Case-First (Top-Down)¶
Start from what AI can do for you today, then peel back the layers. Best if you want to be productive fast and deepen understanding as you go.
The two paths converge after Module 2
Approaches A and B are two different on-ramps: A starts from the science, B from the use cases. They differ only in Modules 1 and 2 — from Module 3 onward (Prompt Engineering, RAG, Agents, Ecosystem) the content is identical. Pick the entry point that suits you; both paths meet at Module 3 and run the same from there.
Both paths share the same Labs, Glossary, and Resources. When you've finished the modules, the Capstone ties everything together — build and evaluate a grounded support agent that retrieves docs and calls a tool.
What you'll be able to do afterward¶
- Explain what an LLM is, how it works, and why it hallucinates — to a non-technical stakeholder.
- Write and iterate on effective prompts for a specific task.
- Explain when to use RAG vs fine-tuning vs just prompting (and when long context or plain code beats all three).
- Design a simple RAG pipeline on a whiteboard.
- Explain what an agent is, what an agent harness owns, and when agentic systems are appropriate.
- Tell a standard model from a reasoning model and choose between them on cost vs. accuracy.
- Navigate the AI framework landscape and recommend the right tool for a use case.
- Identify cost, latency, reliability, and responsible-AI concerns when using LLMs in production.
- Write a basic eval to check an AI feature — and know why "looks right" isn't a passing grade.