AI&DATA SCIENCE

GEN AI, RAG Vector Databases, AGENTIC AI
Generative AI & LLMs & LVMs
• Introduction to Gen AI and LLMs
• Architecture Overview
• Limitations and challenges
• Configuration Parameters for LLMs

LLMs:
• Open AI, Gemini, Llama, Mistral, QWEN…
• Working with text, Images, Video
• Creating own Chatbots
• Working wit databases (SQL function calling)
• evaluation metrics for LLMs (BLEU, ROUGE, perplexity)

Prompt Engineering:
• Prompt Engineering Introduction
• Prompt Engineering and Types
• Few-shots, Zero-shot, Chain-of-Thoughts, Instructional Prompting
• Role-Playing and Open-ended Prompting

Ollama :
• Ollama Overview – What is Ollama and Advantages
• Ollama Model Parameters
• Ollama CLI Commands -Pull and Testing a Model
• Pull in the LLaVA Multimodal Model and Caption an Image
• Summarization and Sentiment Analysis
• Different Ways to Interact with Ollama Models
• Interact with Llama3 in Python Using Ollama REST API

Vector Databases and RAG
• RAG Introduction
• RAG Key Components,
• Introduction to Vector Databases Overview
• Why Vector Databases, Vector Databases – Benefits and Advantages
• Traditional vs. Vector Databases – Limitations and challenges
• Building Vector Databases
• Chroma Database workflow
• Creating a Chroma DB and Adding Documents and Querying
• Looping Through the Results & Showing Similarity Search Results
• Chroma Vector Database – Persisting Data and Saving
• Creating OpenAI Embeddings – Raw without Chroma
• Using OpenAI Embedding API to Create Embedding in Chroma DB
• Vector Databases Metrics and Data Structures
• Vector Similarity Cosine Similarity
• Euclidean Distance – L2 Norm, Dot Product
• Generating Embeddings from Documents and Insert to Vector Database
• Conversation between Llama/Gemini with Notepad text, PDF, Web data.

Agentic AI Frameworks
• AI Agents, Agentic AI
• AI Agentic Design Patterns
LangChain:
• LangChain Fundamentals
• Main Components
• ChatModel,ChatPromptTemplates
• Indexes, Retrievers and Data Preparation
• LangChain TextLoaders
• Text Splitting and Cleaning
• Embeddings and Retriever with FAISS VectorStore
• TextSplitter, DirectoryLoader
• PDFLoader, Chains
• RAG System with Chat and LangChain Chains
• RAG System QA Bot Using LangChain
• Query Expansion with Multiple Queries – Overview
• Query Expansion Full RAG Workflow
• Re-Ranking & Cross-encoder and Bi-encoders – Overview
• Reranking Technique RAG System Workflow Architecture
LangGraph:
• LangGraph Overview, Key Concepts
• Core Concepts – Flow Diagram
• Data and State
• Bot – Streaming Values
• Adding Tools to our Basic LangGraph Agent
• Adding Memory to Our Agent State
• Adding Human-in-the-loop to the AI Agent
• Adding Nodes and Edges and Running our Agent
• Creating REAT agent using LangGraph
CrewAI
• What is CrewAI? Why multi-agent?
• Configure your LLM backend (OpenAI, Groq, or Ollama).
• Agents & Tools
• Learn how to define agents with:
o Role, goal, backstory
o Assign tools (e.g., calculator, web search, custom Python functions).
• Hands-on: Create two agents (e.g., Researcher + Writer).
• Tasks & Crews
o Learn orchestration models:
Sequential (linear steps)
Hierarchical (supervisor assigns tasks)
• Advanced Features
• Custom tools (e.g., Python functions, database queries).
• Human-in-the-Loop (review before final output).
o Implementing RAG CREW
Model Context Protocol:
• MCP overview & history
• Purpose: standardizing model ↔ tool communication
• MCP vs API vs Lang Chain/LLM tool calling
• MCP Architecture
• Build a simple MCP tool (e.g., math_calculator).
• Connect it to a LangChain agent (via custom tool).
• Test LLM → MCP tool call → response.
o Hands-On case studies
• Capstone project on Gen AI/ Agentic AI