By the end
What you'll build
- Explain the think-act-observe agent loop and implement it by hand, including the final-answer and step-cap brakes.
- Apply ReAct and up-front planning appropriately, and decompose goals into deterministically checkable sub-tasks.
- Design agent memory - working, episodic, and semantic - using sliding windows and summary buffers.
- Build multi-tool agents over a mock CRM, calculator, and document corpus that are graded on final environment state.
- Diagnose agent failures from traces, including corrupt success and prompt injection, and classify them precisely.
- Implement guardrails, tool allowlists, retries with backoff, and idempotency to make tool use safe.
- Read the patterns behind LangGraph, CrewAI, and the OpenAI Agents SDK without depending on any single framework.
- Write state-based evaluation assertions instead of judging an agent by its prose.
- Deploy a real agent from an official quickstart and manage keys, cost, and observability responsibly.
- Present sandbox and deployed work honestly in an agent-engineering interview, with no reliance on job or salary promises.
The shape of it
How this course works
Short lessons
44 lessons across 8 modules, each small enough to finish in one sitting.
Practice as you go
Every lesson ends with a small space for what you noticed — the doing is the learning.
A certificate at the end
Finish the course and earn a certificate anyone can verify with a link.
Ready when you are.
Make an account and this course opens up — your progress is saved from the very first lesson.