Course Outline:
Foundational Python for Agentic AI
- Pydantic Library
- Structured inputs/outputs for tools, agents, and workflows
- Typed schemas for message passing
- Models, fields, validators and using BaseModel
- AsyncIO Fundamentals
- Event loops, tasks, coroutines, and concurrency primitives
- Running parallel tool calls, API requests
Containerization With Docker
- Introduction to Containers
- Docker overview
- Docker commands
- Understanding Dockerfile
- Building Docker Containers
- Using Docker-Compose for building and testing software
AI Engineering & Vibe Coding
- Understanding AI in Software Development
- AI’s role in modern programming workflows
- Strengths and Limitations of AI-powered coding tools
- Exploring Free AI Coding Tools
- GitHub Copilot: Auto completion & suggestions
- Codeium: Full AI coding assistant
- Tabnine: AI-assisted code predictions
- Google Gemini for code: AI-generated coding assistance
Pydantic AI: Pythonic Agent Framework
- Schema-First Agents: Defining strict I/O using Pydantic Models to ensure JSON reliability
- Dependency Injection: Injecting database connections and API clients into agents securely
LangChain & LangGraph: Stateful Intelligence
- LangChain Essentials
- Prompt templates
- LCEL (LangChain Expression Language)
- Tool-calling loops
- LangGraph Core Concepts
- Nodes & Edges: Building state machines instead of linear chains
- State Management: Using reducers to merge agent updates into a “source of truth”
- Persistence & Checkpointing: Implementing SQLite/Postgres savers to pause and resume agent tasks
- Human-in-loop: Designing nodes that wait for human approval before executing sensitive API calls
- LangSmith
- Debugging complex traces
- Running automate eval-frameworks
Model Context Protocol (MCP)
- The Protocol Spec
- Understanding the Client-Server relationship in MCP
- STDIO and Streamable HTTP
- Pre-built Servers
- Connecting to GitHub, Google Drive and Postgres
- Building Custom MCP Servers
- Creating Python/Node.js servers to expose internal legacy APIs to LLMs
- Secure Resource Management: controlling file/data that the AI can “see”
Automation & Low Code Orchestration
- Multi-Agent n8n Workflows: building orchestrators that manage multiple sub-agents
- Self-Healing Workflows: Implementing error-handling loops that ask an LLM to “fix” a failed HTTP request
- Self-Hosting n8n: Deploying via Docker for enterprise data sovereignty
Enterprise RAG & Database Integration
- Advanced Vector Search: Hybrid search (Keyword + Semantic), reranking, and parent-document.
- Database bridges: Using SQL Agents to turn natural language into secure, read-only Postgres queries
Local AI, Privacy & Security
- Local LLM Deployment: Running Ollama, DeepSeek, and Qwen via Docker
- The AI Security Stack
- Guardrails: Using Guardrails to prevent toxic or off-topic outputs
- Compliance: Mapping agent behavior to GDPR and the EU AI Act
- Prompt Injection Defense: Techniques for sanitizing user inputs in agentic workflows
Multi-Modal Agents & Deployment
- Voice Agents: Low-latency architecture for creating voice agents
- Generative Pipelines: Using tools to create visual assets
- Cloud hosting: Containerizing agents with Docker and deploying them to cloud
Capstone Project & Use Case
- Project Overview
- Complete projects to get experience and practice
- Presentation of project findings and Insights
- Industry Use Case Studies
Certification (Optional)
- Certification Overview
- Identify the right certification for you
- Tips to prepare for certification
