Agentic AI

Get production-ready with the Python and GenAI foundations every agent builder needs.

• Advanced Python (typing, dataclasses, decorators)
• Async & concurrency with asyncio
• Virtual environments & dependency management
• Building & consuming REST APIs with FastAPI
• Data validation with Pydantic
• HTTP clients, retries & rate limiting
• Intro to GenAI & the LLM landscape
• Tokens, context windows & pricing
• Calling LLM APIs (OpenAI / Anthropic)
• Environment & secrets management
• Git & collaborative workflows

Engineer reliable prompts and choose the right model for each job.

• Prompt anatomy & design patterns
• Zero-, one- & few-shot prompting
• Chain-of-Thought & reasoning prompts
• Role & system prompting
• Structured output (JSON mode, schemas)
• Prompt templates & variables
• Guardrails & prompt-injection defense
• LLM comparison & selection (GPT, Claude, Llama, Gemini)
• Temperature, top-p & sampling controls
• Token cost & latency optimization
• Evaluating prompt quality

Understand how autonomous agents reason, plan, and act.

• What is an AI agent? The agent loop
• Agent architectures (reflex, goal-based, utility-based)
• The ReAct framework (reason + act)
• Planning & task decomposition
• Reflection & self-critique
• Tool use & function-calling fundamentals
• Agent memory basics
• Agent evaluation & benchmarking
• Failure modes & guardrails
• Cost & latency of agent loops
• Design patterns for reliable agents

Master the industry-standard frameworks for building agentic apps.

• LangChain core concepts
• LCEL & Runnables
• Chains & composition
• Prompt & output parsers
• Tools & toolkits
• LangGraph state machines
• Nodes, edges & conditional routing
• Cyclic graphs & loops
• Persistence & checkpointing
• Streaming & async execution
• LangSmith tracing & debugging
• Human-in-the-loop patterns

Orchestrate teams of specialized agents that collaborate.

• Multi-agent system (MAS) fundamentals
• Agent roles & specialization
• Orchestration patterns (hierarchical, sequential, collaborative)
• Agent-to-agent communication
• Function / tool calling in depth
• Building custom tools
• CrewAI crews & tasks
• AutoGen conversational agents
• Model Context Protocol (MCP): servers, clients, tools
• Shared state & message passing
• Negotiation & consensus mechanisms
• Debugging multi-agent flows

Ground agents in your data with retrieval-augmented generation.

• RAG architecture & when to use it
• Document loaders & parsing (PDF, HTML, docs)
• Chunking & text-splitting strategies
• Embeddings & embedding models
• Vector databases (Pinecone, Chroma, Weaviate)
• Similarity search & indexing
• Hybrid keyword + semantic retrieval
• Re-ranking & retrieval quality
• RAG evaluation & fallback strategies
• Metadata filtering
• Multi-document & agentic RAG
• Reducing hallucinations

Give agents short- and long-term memory and let them run autonomously.

• Conversation & short-term memory
• Long-term & persistent memory
• Vector-backed memory stores
• Summarization & context compression
• Autonomous task-execution loops
• Planning & re-planning
• Error recovery & retries
• Self-healing agents
• State persistence & checkpoints
• Budget & step limits
• Observability of long-running agents

Wire agents into real systems, events, and third-party APIs.

• Workflow orchestration patterns
• Event-driven & triggered agents
• Webhooks & callbacks
• Integrating external APIs & SaaS tools
• Authentication & secrets handling
• Rate limiting & retries
• Queues & background jobs
• Scheduling & cron agents
• Streaming responses to clients
• Structured tool outputs
• Idempotency & reliability

Ship a production copilot with a UI, deployment, and monitoring.

• Copilot architecture & UX patterns
• Building UIs with Streamlit & Gradio
• Chat interfaces & streaming
• Packaging with Docker
• Deploying to the cloud (containers)
• Environment config & secrets
• Observability & tracing (LangSmith)
• Logging, metrics & cost monitoring
• Evaluation & regression testing
• Guardrails & safety in production
• Capstone: end-to-end AI copilot