AI Agents Engineering
Build agents that finish the job
Multi-agent systems, memory, protocols, and observability — the deep engineering of reliable autonomous AI.
Advanced3 modules · 10 lessons~104 min1060 XPAligned with Hugging Face Agents CourseStart with: Applied GenAI
Progress0/10
Module
Agent Internals
The loop under load, planning under uncertainty, memory that actually works
Inside the Agent Loop10 min · 100 XPInside the Agent Loop
Planning Strategies11 min · 110 XPPlanning Strategies
Memory & State11 min · 110 XPMemory & State
Module
Tools & Protocols
Designing tools agents survive, MCP standardization, and agents that drive the screen
Tools Agents Can Actually Use10 min · 100 XPTools Agents Can Actually Use
MCP: The Model Context Protocol10 min · 100 XPMCP: The Model Context Protocol
Computer-Use Agents11 min · 110 XPComputer-Use Agents
Module
Multi-Agent & Production
When more agents help, how to see inside a run, and autonomy that earns trust
Multi-Agent Patterns11 min · 110 XPMulti-Agent Patterns
Agent Observability10 min · 100 XPAgent Observability
Agent Evals11 min · 120 XPAgent Evals
The Autonomy Dial9 min · 100 XPThe Autonomy Dial
Capstone project
Ship an agent
Design (and optionally build) a small agent: its tools, its planning loop, and — critically — where it stops and asks a human. Test three tasks it should refuse to guess on.
Deliverable: An agent design doc or working demo including the escalation/fallback logic.
Milestones
Finish every lesson to earn the AI Agents Engineering Certificate.