5. AI Agent Development
Construct intelligent, autonomous systems using state-of-the-art agent architectures, structured JSON schemas, function-calling tool registries, and stateful memory loops.
Mario Guerra
Author & Technical Leader
Course Outline
Orientation: AI Agent Development
Welcome to AI Agent Development. Set up your environment and obtain a free Gemini API key.
System Instructions & Boundaries
Control model behavior and define strict execution boundaries using System Instructions.
Structured Outputs (Pydantic)
Force LLMs to return strict JSON data matching Pydantic schemas to prevent parsing crashes.
Tool Calling & Function Execution
Register local Python functions as tools to allow Gemini to read and write database files autonomously.
Stateful Memory Serialization
Build a sliding-window conversation history manager and serialize message states to disk.
Multi-Agent Orchestration & HITL
Coordinate a Planner Agent and Auditor Agent using a Triage Router with Human-in-the-Loop approval gates.
Epilogue: Portfolio Graduation
Review your completed distributed system, construct resume bullet points, and prepare to demo your project to recruiters.