Engineer AI agents that work in production.

No. Phase 1 starts from setup and fundamentals, so you do not need machine-learning theory, model training, or previous AI project experience. What helps is comfort with basic programming—variables, functions, loops—and using a terminal.

Basic Python, or another language you can pick up quickly, plus a willingness to read documentation. Git, APIs, and command-line work appear early in Phase 1, and every module includes hands-on practice, so gaps get closed as you go rather than being assumed away.

Plan for around 6–8 hours a week: short lessons plus the hands-on build for that module. Learners who can only give a few hours a week still finish, but the capstone benefits from a longer, uninterrupted stretch near the end.

Most learners move through all three phases in roughly 12–16 weeks at a steady pace. The phases build in order—foundations, then orchestration, then production and the capstone—so the sequence matters more than the calendar.

Enroll from the course page; every Learn now button on this site opens it. For a full breakdown of what each phase includes and the ways to pay, see the pricing & enrollment page. The syllabus details every module before you decide.

The phases are ordered on purpose. Phase 2 assumes you can already build and call an application, retrieve context, and work with prompts. Phase 3 assumes you can orchestrate stateful workflows and evaluate them. If you already work with agents, the Phase 1 and Phase 2 module lists are the quickest way to check your gaps.

Six modules that turn setup into working software: a Python and AI development environment, APIs and chat interfaces with FastAPI, Streamlit, and Gradio, LLM fundamentals and prompt engineering, embeddings and vector search, AI-assisted coding practice, and no-code workflow automation. You finish with several small applications you built yourself.

Eight modules on turning isolated model calls into systems: LangChain chains, memory, and RAG; agents with typed tools; LangGraph workflows with routing, cycles, checkpoints, and human review; tracing, evaluation, and testing; the Model Context Protocol with custom MCP servers; and programmatic prompting. You finish with an orchestrated agent you can measure.

Eleven modules on making it real: deep agents with planning and long-term memory, multi-agent orchestration, agentic RAG and GraphRAG, visual automation, guardrails and safety, monitoring and fine-tuning, containerizing and deployment, agent interoperability, and LLM security. It closes with a two-part capstone you design, build, deploy, and document.