Agentic AI

Course Overview

Foundations of Agentic AI and LLMs

• Introduction to Agentic AI

• Agentic AI vs. Generative AI

• Types of agents: reactive, proactive, collaborative

• LLM architecture: tokens, embeddings, transformers

• Setting up your development environment (Python, OpenAI API)

• Mini Project: Build your first conversational AI agent

Conversational Pipelines and RAG (Retrieval-Augmented Generation)

• NLP pipelines and conversation design

• Prompt engineering and prompt chaining

• Embeddings, vector databases (FAISS, Pinecone)

• Retrieval-based augmentation of LLM outputs

• Project: Create a Q&A system over documents using RAG

LangChain Fundamentals & Tool Integration

• Understanding chains, memory, tools, and agents in LangChain

• Integrating external APIs and functions

• Structured outputs with output parsers

• Prompt templates and memory stores

• Project: Build a LangChain-based productivity assistant

LangGraph and Stateful Agent Workflows

• Introduction to LangGraph for multi-step workflows

• Managing agent state, transitions, and memory

• Error handling and response strategies

• Project: Design a finance assistant with LangGraph workflows

Multi-Agent Systems with CrewAI and AutoGen

• Multi-agent collaboration frameworks

• Role-based delegation and task management (CrewAI)

• Building reflexive agents using AutoGen

• Agent feedback loops, self-evaluation, and decision logic

• Project: Implement a writing & editing agent team

Deployment and Hosting of Agentic Systems

• Introduction to OpenAI Agents SDK & function calling

• Hosting agents with FastAPI and Streamlit

• Cloud integrations (AWS, GCP, Azure)

• Monitoring and observability (LangFuse, Portkey)

• Project: Deploy and test your personal AI assistant online

Security, Ethics, and Trust in Agentic Systems

• TRiSM framework (Trust, Risk, Security Management)

• Data privacy and model security

• Handling hallucinations, bias, and misuse

• Ethical agent design and governance

• Assignment: Perform a risk audit for your deployed agent

Capstone Project & Future Directions

• Capstone Project: Design, implement, and demo a complete agent system (e.g., Legal researcher, CRM assistant, AI tutor)

• Advanced concepts: self-evolving agents, multimodal agentic AI, long-term memory

• Career pathways: Agent Developer, AI Workflow Engineer, LLMOps Specialist