Generative AI Course Curriculum
1. Basics of GenAI
Generative AI is transforming the technological landscape by enabling machines to create text, code, and media. This course introduces the foundational concepts of Large Language Models (LLMs) and foundation models. Students will learn the underlying architecture of generative models, effective prompt engineering techniques, and how to securely interact with industry-leading AI APIs. Through practical exercises, learners gain an intuitive understanding of model capabilities, limitations, and the ethical guardrails required to develop safe and responsible generative applications.
Introduction to Generative AI
- Evolution of AI vs Generative AI
- Foundation Models & LLMs
- Generative vs Discriminative Models
- Use Cases & Applications
Prompt Engineering Fundamentals
- Zero-shot & Few-shot Prompting
- Chain of Thought Reasoning
- Prompt Optimization Techniques
- Managing Context Windows
Working with LLM APIs
- OpenAI & Anthropic APIs
- Temperature & Top-P Sampling
- Tokenization Fundamentals
- Handling API Responses & Streaming
Ethics, Safety & Evaluation
- Hallucinations & Mitigation
- Bias, Toxicity & Safety Guardrails
- Model Evaluation Metrics
- Data Privacy & Security
2. RAG Pipelines
Learn to build Retrieval-Augmented Generation (RAG) systems that ground Large Language Models in external knowledge bases. This course covers the end-to-end pipeline required to reduce hallucinations and make AI applications context-aware. You will gain hands-on experience with vector databases, embedding models, document chunking, and advanced retrieval strategies. By the end of the course, you will be able to construct robust architectures capable of querying proprietary data and delivering highly accurate AI responses.
RAG Fundamentals
- What is RAG?
- Embeddings & Vector Spaces
- Cosine Similarity
- The RAG Architecture
Data Ingestion & Processing
- Document Loaders
- Text Splitters & Chunking Strategies
- Metadata Extraction
- Handling Unstructured Data
Vector Databases
- Vector DB Concepts
- Indexing & Semantic Search
- Hybrid Search (Keyword + Vector)
- Pinecone, Chroma & Milvus
Advanced RAG Techniques
- Query Routing & Expansion
- Re-ranking Models
- Context Synthesis
- Evaluating RAG with RAGAS
3. MCP (Model Context Protocol)
Master the Model Context Protocol (MCP) to seamlessly connect AI models with external tools, APIs, and real-time data sources. This course teaches the standardized architecture required to implement reliable context windows and orchestrate secure client-server model interactions. Students will learn how to structure tool calls, manage context lifecycle, and build interoperable AI integrations that allow language models to securely access and reason over live enterprise environments.
Introduction to MCP
- What is Model Context Protocol?
- The Client-Server Architecture
- Standardizing Tool Calling
- Use Cases for MCP Interoperability
Building MCP Servers
- Defining Server Resources
- Implementing Standardized Prompts
- Creating Tool Schemas
- Authentication & Security Measures
Client Integration
- Connecting LLMs to MCP Servers
- Parsing Tool Execution Requests
- Handling Tool Responses
- Error Handling & Retries
Advanced MCP Implementation
- Dynamic Context Loading
- Managing Context Lifecycle & Token Limits
- Multi-Server Orchestration
- Enterprise Data Integration
4. Agentic AI Development
Transition from basic conversational AI to autonomous AI agents capable of planning, reasoning, and executing multi-step workflows. This course covers cognitive agent architectures, persistent memory systems, and tool orchestration. Students will use modern frameworks to design, deploy, and monitor multi-agent systems that can autonomously solve complex problems, interact with software environments, and collaborate effectively with human oversight.
Fundamentals of AI Agents
- Traditional AI vs Agentic AI
- Single-Agent vs Multi-Agent Systems
- Agent Lifecycle and Execution Flow
- Common Agent Frameworks
Building Your First AI Agent
- Setting Up Development Environment
- Connecting to LLM Models
- Prompt Templates & System Instructions
- Creating Reasoning Capabilities
- Tool Calling & Function Invocation
- Error Handling, Retries & Fallbacks
Memory & External Tools
- Introduction to Agent Memory
- Short-Term vs Long-Term Conversation Memory
- Web Search & Calculator Tools
- Database Query Integration
- Building Custom Agent Tools
Enterprise Workflows & Best Practices
- Multi-Step Task Execution
- Planner & Worker Agents
- Linear & Multi-Agent Orchestration Patterns
- Agent Planning, Reflection & Loops
- Guardrails & Responsible AI
- Observability, Logging, Debugging & Tracing
5. Fine Tuning LLM
Take control of Large Language Models by adapting them to specialized domains and custom tasks through Fine Tuning. This course explores the mechanics of Supervised Fine Tuning (SFT) and modern, parameter-efficient techniques like LoRA and QLoRA. Gain practical expertise in dataset preparation, optimizing compute resources on GPUs, and evaluating trained models. You will learn to successfully tailor open-source models to specific business needs while preserving their core reasoning capabilities.
Fine Tuning Fundamentals
- Pre-training vs Fine Tuning
- When to Fine Tune vs Use RAG
- Supervised Fine Tuning (SFT)
- Transfer Learning Concepts
Dataset Preparation
- Data Formatting (JSONL, ChatML)
- Instruction Tuning Datasets
- Data Cleaning & Validation
- Tokenization Strategies
Parameter-Efficient Fine Tuning (PEFT)
- LoRA (Low-Rank Adaptation)
- QLoRA & Quantization
- Freezing Model Weights
- Optimizing GPU Compute Resources
Training & Deployment
- Training Loops with Hugging Face
- Checkpointing & Evaluation
- Merging Adapters with Base Models
- Deploying Custom Models
