AI 1002 · Agentic AI Engineering

AI 1002

AI 1002 ·middot; Six modules · 12 tutorials · requires Python

Building production AI agent systems — from API integration and tool use through four major frameworks to the Model Context Protocol. Each module covers a different approach to agent architecture: the OpenAI Agents SDK for structured agent development, Crew AI for collaborative agent teams, LangGraph for graph-based workflows, and AutoGen for distributed systems. The course culminates in MCP — the protocol that makes all agent frameworks interoperable.

Prerequisites: Python programming ability. Familiarity with async/await. API experience helpful but not required. AI 1001 (AI for Beginners) recommended but not mandatory.

Code: AI 1002Level: IntermediateProvider: Universitas Scholarium

Module 1Foundations of Agent Engineering2 tutorials

Hosted by the Marvin Minsky Simulacrum

Connecting to LLM APIs · tool use and function calling · multi-model orchestration · five essential workflow design patterns · the agent loop · deploying to Gradio.

Open module →

Module 2The OpenAI Agents SDK2 tutorials

Drawing on the engineering insights of Ilya Sutskever

Async Python · Agent, Runner and Trace classes · hierarchical agents · guardrails and structured outputs · deep research agents with parallel search · Gradio deployment.

Open module →

Module 3Collaborative Agent Teams2 tutorials

Drawing on the multi-agent research methodology of Demis Hassabis

Crew AI framework · agents, tasks and processing modes · custom tools · web search · vector memory · code-generating agents · multi-agent financial research.

Open module →

Module 4Graph-Based Agent Architecture2 tutorials

Hosted by the Edsger Dijkstra Simulacrum

LangGraph · nodes, edges and state · conditional routing · checkpointing with SQLite · Playwright browser automation · worker-evaluator feedback loops · AI sidekick with memory.

Open module →

Module 5Distributed Agent Systems2 tutorials

Hosted by the Claude Shannon Simulacrum

Microsoft AutoGen · agent chat with tools · multimodal features · message handlers and dispatching · distributed runtime with gRPC · autonomous self-replicating agents.

Open module →

Module 6The Model Context Protocol2 tutorials

Hosted by the Alan Turing Simulacrum

MCP architecture (hosts, clients, servers) · building custom MCP servers · external API integration · autonomous trading agents · framework selection · 10 essential lessons for production agent systems.

Open module →