Generative AI Using Python: Building Production Ready Intelligent Applications

Introduction to Generative AI Ecosystem

• Evolution of Generative AI and foundation models
• Architecture of Large Language Models
• Local vs cloud-based inference patterns
• OpenAI ecosystem overview
• Open-source model landscape (Mistral, LLaMA, Falcon)
• Cost-performance considerations

Architecture discussion:

• SaaS APIs vs self-hosted models
• Build vs integrate decision framework

Hands-on:

• Configure OpenAI API
• Run local inference using Ollama

Prompt Engineering & Design Patterns

• Role-based prompting strategies
• Structured PromptTemplates
• Context injection and dynamic inputs
• Output formatting and schema enforcement
• Temperature, top-p, and response control
• Prompt versioning and testing strategies

Hands-on:

• Build Travel Guidance application
• Develop Interview Preparation assistant using LangChain

LangChain Expression Language (LCEL) & Chain Logic

• Introduction to LangChain architecture
• LCEL fundamentals
• Sequential chains and branching logic
• Multi-model integration patterns
• Passing intermediate outputs across steps
• Structured outputs for automation

Hands-on:

• Build automated Blog Post Generator
• Develop Marketing Content Engine

Conversational AI & Memory Management

• ChatMessageHistory fundamentals
• StreamlitChatMessageHistory integration
• Short-term vs long-term memory strategies
• Session persistence and state management
• Designing context-aware applications

Hands-on:

• Build persistent conversational chatbot
• Implement session-based memory logic

Embeddings & Vector Database Architectures

• Understanding embeddings and semantic similarity
• Text chunking strategies
• Indexing and retrieval optimization
• FAISS implementation fundamentals
• Comparison of embedding models
• Performance and storage considerations

Hands-on:

• Build semantic search system
• Optimize chunk size and retrieval accuracy

Real-world application:

• Designing internal knowledge search systems

Retrieval-Augmented Generation (RAG) Pipelines

• RAG architecture overview
• Document ingestion and indexing
• Retriever–Generator integration
• Handling complex PDFs and structured documents
• Hallucination mitigation techniques
• Debugging retrieval bottlenecks

Hands-on:

• Build document-aware RAG bot
• Develop Legal Document Analysis assistant

Real-world application:

• Secure enterprise document intelligence platform

Building AI Web Applications with Streamlit

• Streamlit fundamentals
• Creating interactive UI components
• Connecting frontend to LLM backends
• Error handling and user validation
• Multi-step reasoning workflows
• Session state management

Hands-on:

• Develop responsive AI web application
• Implement guardrails and validation checks

Real-world application:

• Deploying AI-powered customer-facing applications

Intelligent Workflow Design & Use Cases

• Image analysis workflows
• KYC automation logic
• Personalized health assistant design
• Multi-step reasoning chains
• Security considerations in AI applications

Hands-on:

• Build multi-step AI workflow application

Agentic AI & Model Context Protocol (MCP)

• Agent fundamentals (tools, memory, reasoning)
• Tool calling and reactive agents
• Multi-agent orchestration patterns
• MCP architecture (server-client model)
• STDIO vs HTTP-based MCP integration
• Designing scalable tool-enabled AI systems

Hands-on:

• Build Landmark Recognition Agent
• Deploy custom MCP server

Capstone Project: Enterprise Generative AI Application

Participants will design and implement a full-stack AI solution integrating:

• Advanced prompt engineering
• RAG pipeline
• Persistent memory
• Tool-based agent architecture
• Streamlit-based UI
• Deployment-ready structure

Recommended Themes:

• Enterprise Knowledge Management System
• Regulatory Compliance Automation Bot
• AI-Driven Research & Synthesis Assistant

Project Deliverables:

• Architecture diagram
• Source code repository
• Demonstration of live system
• Performance and scalability discussion