AI-Assisted Programming Advanced

Module 1

Building your own AI Agent

Learn how AI agents are designed, assembled, and evaluated in real engineering environments. In this module, you'll build a complete agent workflow, exploring how tools, memory, context, and human oversight work together to support reliable task execution.

• Define the scope, objectives, and success criteria for an AI agent
• Prepare and connect data sources, tools, and resources for agent use
• Build agent workflows using Claude Code and LangGraph
• Add memory and supporting functionality to improve performance
• Apply safety, security, and human review mechanisms to agent workflows

• Claude Code
• LangGraph
• Playwright
• LangSmith
• OpenTelemetry

QA Agent Project Build a QA agent that reads engineering tickets, enriches context from a knowledge base, generates test-case scenarios, routes outputs through a human review process, and publishes approved results back to the ticket.

Module 2

Extending AI with custom MCP servers

Explore how MCP enables AI systems to access organizational knowledge, tools, and data sources in a structured and secure way. You'll learn how MCP servers extend agent capabilities and support integration with real-world engineering environments.

• Explain the role of MCP in AI systems and evaluate existing MCP solutions
• Design MCP interfaces that connect data sources to tools and resources
• Implement and maintain MCP server handlers
• Validate, test, and debug MCP server implementations
• Containerize MCP services using Docker
• Integrate MCP servers into agent workflows and development environments

• MCP / MCP SDK
• Python
• MCP Inspector
• Docker
• pytest
• Claude Code
• Cursor

Git Activity Analyzer MCP Server Build an MCP server that exposes repository activity data—including commit patterns, file hotspots, build history, team structure, and code ownership—as resources, tools, and prompts for AI agents.

Module 3

Orchestrating multi-agent systems

Learn how complex engineering workflows can be coordinated through multiple specialized agents. This module introduces multi-agent architectures, orchestration patterns, and stateful workflows that enable agents to collaborate, and operate within controlled execution environments.

• Identify when multi-agent architectures are more effective than single-agent systems
• Map engineering workflows to appropriate orchestration patterns
• Define specialized agent roles with clear responsibilities and interfaces
• Build stateful multi-agent systems using LangGraph
• Manage shared state and coordination across agents
• Implement conditional routing, human-in-the-loop checkpoints, and execution tracing with LangSmith

• LangGraph
• LangSmith

Multi-Agent Code Review System Build a multi-agent code reviewer composed of specialized agents for bug detection, style review, code improvement, and summarization. The system includes conditional routing, retry mechanisms, observability with LangSmith, and human approval checkpoints. Guided cohorts may also apply the same architecture patterns to a custom project.