What You'll Build: Semantic search engines, PDF QA systems, conversational agents, multi-agent workflows, and a fully deployed production application with observability and safety mechanisms.

Bonus: Explore OpenAI's ChatGPT Agent Kit for building custom GPT agents.

Perfect for engineers ready to lead GenAI initiatives in their organizations.

GenAI Engineering Training Curriculum

Week 1: GenAI Foundations

Session 1: Introduction to Generative AI

• Evolution from AI → ML → DL → GenAI

• Core components: LLMs, Embeddings, Vector DBs, RAG, Agents

• Industry use cases in healthcare, legal, and retail

• Open vs closed source LLMs

Session 2: Prompt Engineering & Embeddings

Prompt Engineering Fundamentals:

• Prompt structure and formatting best practices

• Zero-shot, Few-shot, Chain-of-Thought techniques

• ReAct, Tree-of-Thought, and prompt tuning strategies

• Common prompt pitfalls and debugging

Text Embeddings & Semantic Similarity:

• Intro to text embeddings and semantic similarity

• Embedding models: OpenAI, Cohere, HuggingFace

• Understanding vector dimensions and inference performance

• Comparing cosine similarity vs Euclidean distance

Week 2: Vector Databases & RAG Basics

Session 3: Vector Databases

• Intro to FAISS, Pinecone, Chroma, Weaviate

• Indexing strategies: Flat, IVF, HNSW

• Vector CRUD operations and metadata filtering

• Performing semantic search over chunked documents

• Assignment: Build a basic semantic search engine using ChromaDB

Session 4: Retrieval-Augmented Generation (RAG) with LangChain

• Understanding the RAG pipeline and architecture

• Chunking, indexing, and context injection

• LangChain retriever + prompt + LLM orchestration

• Template design for context-aware prompts

• Assignment: Build a PDF QA app using RAG and LangChain

Week 3: RAG Frameworks

Session 5: RAG with Phi Data / Agno AGI

• Modular RAG workflows using Phi Data or Agno AGI

• Handling complex inputs and retrieval logic

• Lightweight integration with LLM APIs

• Combining structured and unstructured sources

Session 6: RAG with Amazon Bedrock

• Using Titan, Claude, and other models via Bedrock

• Setting up and securing a RAG workflow in Bedrock

• Integration with S3, RDS, or DynamoDB

• Monitoring cost and performance

Week 4: Advanced RAG

Session 7: RAG Workflow with LangFlow

• LangFlow overview and no-code chaining

• Node types: retriever, prompt, LLM, output

• Combining structured and unstructured tools

• Exporting and deploying LangFlow project

Session 8: Advanced RAG Techniques

• Advanced chunking strategies and document preprocessing

• Hybrid search: combining semantic and keyword search

• Query transformation and routing

• Context compression and reranking

• Assignment: Implement hybrid search with reranking in a RAG pipeline

Week 5: Agentic AI

Session 9: Agentic AI Concepts + LangChain Agents

• What is Agentic AI? Tools, Memory, Planning

• LangChain agent executor and agent types

• Connecting LLMs with tools (API, calculator, search)

• Adding memory to track multi-turn context

Session 10: Build Agents with LangFlow

• Using LangFlow to wire agents visually

• Creating tools and chaining agent steps

• Building a multi-turn conversational agent

• Exporting and reusing LangFlow configs

Week 6: Agent Orchestration & Protocols

Session 11: Multi-Agent Workflows with CrewAI

• CrewAI: roles, tasks, and collaborative agents

• Assigning memory and tools to agents

• Simulating agent-based research and review

• Evaluating multi-agent output quality

Session 12: MCP and A2A Protocols

Model Context Protocol (MCP):

• Understanding MCP architecture and components

• Connecting AI models to external data sources

• Building MCP servers and clients

• Use cases: database integration, API connections, tool access

Agent-to-Agent (A2A) Protocols:

• Inter-agent communication standards

• Message passing and coordination patterns

• Protocol design for multi-agent systems

• Integration with existing agent frameworks

Week 7: Observability & Evaluation

Session 13: Observability with LangChain + LangSmith

• Logging prompt input/output, latency, and feedback

• Tracing tool and memory steps in LangChain apps

• Evaluating hallucinations and grounding

• LangSmith dashboard walkthrough

Session 14: LangWatch for Evaluation & Feedback Loops

• LangWatch vs LangSmith comparison

• Live monitoring, prompt scoring, and user feedback

• Custom evaluation functions (RAGAS, MT-Bench)

• Integrating with continuous improvement loop

Week 8: Production Readiness & Capstone

Session 15: Deployment Strategies & Guardrails

Deployment Options:

• Containerization with Docker

• Cloud deployment: AWS, Azure, GCP

• Serverless architectures (Lambda, Cloud Functions)

• API gateway setup and management

Scaling & Performance:

• Load balancing and auto-scaling

• Caching strategies for LLM responses

• Rate limiting and quota management

• Cost optimization techniques

AI Safety Mechanisms:

• Content filtering and moderation

• Prompt injection prevention

• Output validation and sanitization

• Implementing Guardrails AI or NeMo Guardrails

Security Best Practices:

• API key management and rotation

• Data privacy and compliance (GDPR, HIPAA)

• Audit logging and access controls

• Testing for adversarial inputs

Session 16: Capstone Project

• Choose a domain: Healthcare, Legal, HR, or Finance

• Build end-to-end GenAI application: Embed → RAG → Agent → Guardrails → Observability

• Prepare comprehensive documentation: Architecture diagram, GitHub repository, deployment guide

• Development, testing, and deployment phase

• Live presentation and demo

• Final deliverables: Deployed application + GitHub repo + walkthrough video

• Peer review and feedback session

• Course completion and certification

Bonus Session: ChatGPT Agent Kit

Session 17: Building with ChatGPT Agent Kit

ChatGPT Agent Kit Overview:

• Introduction to OpenAI's Agent Kit framework

• Understanding the Agent Kit architecture

• Pre-built components and templates

Building Custom GPTs:

• Creating custom actions and functions

• Integrating external APIs and tools

• Managing conversation context and memory

Advanced Agent Patterns:

• Multi-step reasoning workflows

• Tool chaining and orchestration

• Error handling and fallback strategies

Production Deployment:

• Publishing custom GPTs

• Monitoring usage and performance

• Iterating based on user feedback

Assignment: Build and deploy a custom GPT agent using ChatGPT Agent Kit