Generative AI

This program offers hands-on experience in building real-world Generative AI products. You will learn to design, develop, and deploy end-to-end AI solutions using industry-leading tools and frameworks. The course covers practical implementation with technologies like HuggingFace Transformers, LangChain, LLaMA, Mistral, Phi, Gemma, OpenAI, Meta SAM, ReAct, AutoGen, CrewAI, and AgentVerse. By mastering modern Generative AI tools, you’ll gain the skills to build complete AI products that include multimodal applications, automated task workflows, and real-time, event-driven agents.

Start your journey to a global career in Generative AI! Gain hands-on skills and become an industry-ready professional

03 Months

Saturday, Sunday

2:00 - 4:00 PM PKT

To be announced

Online & Interactive

English

What You’ll Learn

Module 1: Foundations of Generative AI & LLMs

Understanding Generative AI: History, Use Cases (text, image, code, audio)
Introduction to Transformer architecture
Open Source LLMs (LLaMA, Mistral, Phi, Gemma)
HuggingFace ecosystem: Models, Datasets, Transformers
Google Colab Setup, Ollama on Mac M1
Hands-on:
- Run your first LLM locally using Ollama
- Explore LLaMA 2 and Mistral via HuggingFace
Project: Generate personalized creative bios using pre-trained LLMs

Module 2: Prompt Engineering & Fine-Tuning Basics

Prompt engineering techniques (zero-shot, few-shot, CoT)
Prompt templates with LangChain & LlamaIndex
Fine-tuning vs. instruction tuning vs. LoRA
PEFT (Parameter Efficient Fine-Tuning) with QLoRA
Hands-on:
- Design effective prompts using PromptLayer
- Fine-tune a small LLM using PEFT on Google Colab
Project: Build a custom Q&A bot with fine-tuned prompts

Module 3: Python for LLM Engineering

Intro to python essentials
Data types, control structures, functions, OOP
NumPy, Pandas, Matplot and other libraries
Data pre-processing, wrangling & scaling
Python tools for AI: transformers, datasets, torch, langchain, openai
Reading/writing structured, semi-structured (JSON, YAML) data
Using API & file I/O for data ingestion
Hands-on:
- Use LangChain to wrap your first LLM pipeline
- Manipulate data for prompt engineering
Project: Build a prompt interface in Python that takes CSV/JSON as input

Module 4: LLM-powered RAG Systems (Retrieval Augmented Generation)

What is RAG and why it matters
Building a RAG pipeline: Document -> Chunk -> Embed -> Retrieve -> Generate
Vector stores: ChromaDB, FAISS
Text chunking & embedding with LangChain
Hands-on:
- Load PDFs or docs, chunk, embed, and retrieve with ChromaDB
- Build a search-augmented chatbot with LangChain
Project: RAG system for company policy documents

Module 5: Agentic Workflows & Tool Use

What are LLM Agents?
LangChain agents, ReAct, Tool usage
Calling APIs from LLMs
Designing custom tools & toolkits
Hands-on:
- Build a LangChain agent that queries tools (e.g., weather, calculator, document lookup)
Project: AI HR Assistant that uses tools to answer user queries

Module 6: Memory, Context Management & Chaining

Short vs long-term memory
Conversation memory with LangChain (Buffer, Summary, Entity memory)
Conversation chain vs sequential chain vs custom logic
Hands-on:
- Implement memory-enhanced chatbot
- Build a sequential chain for onboarding workflow
Project: Multi-step agent with memory for employee onboarding

Module 7: Vector Databases & Knowledge Base Integration

Introduction to Vector Search Concepts
Embedding models (OpenAI, Sentence Transformers, InstructorXL)
Working with ChromaDB, FAISS
Knowledge base ingestion & chunking strategies
Hands-on:
- Build document-to-vector pipeline
- Query multiple sources using hybrid search
Project: Internal company knowledge base chatbot

Module 8: Real-Time Data & Event-Driven Agents

Real-time sources: APIs, streams, sensors, events
Agents + streaming workflows (LangChain + Kafka/FastAPI)
Building live alert systems (e.g., anomaly detector with alerts)
Hands-on:
- Use Kafka/Socket API to push live data to agent
- Trigger LLM response on real-time event
Project: Live incident response agent for IT operations

Module 9: Transformers, Fine-Tuning & Vision Models

Deep dive into Transformer architecture
Fine-tuning strategies and use cases
Vision Transformers (ViT) and Segment Anything Model (SAM)
Hands-on:
- Fine-tune a Transformer model on a custom dataset
- Segment images using Meta’s SAM
Project: Vision Transformer app for image labeling/classification

Module 10: AI Agents + Task Automation Platforms

Multi-agent collaboration basics
Introduction to AutoGen, CrewAI, AgentVerse
Task delegation and chaining with agents
Hands-on:
- Build an automated assistant using CrewAI or AutoGen
- Coordinate tasks among agents
Project: Autonomous agent pipeline for customer support or operations

Module 11: Generative AI for Multimodal Applications

Intro to image, audio, and code generation
Text-to-image with Stable Diffusion, DALL·E, Midjourney
Whisper for speech-to-text
AudioLM, MusicGen basics
Hands-on:
- Create a text-to-image app using HuggingFace Diffusers
- Build a voice interface using Whisper + LLM
Project: Voice-enabled AI storytelling app

Tools you will learn

Meet your Instructor

Tayyaba is a Senior AI Engineer with over 4 years of experience in designing and optimizing global ETL, DA, Gen AI, and Agentic AI solutions. She has strong hands-on expertise in AI, RAG, Explainable AI, and managing end-to-end Real-Time Agentic AI Projects.

At Afiniti, I built projects from scratch, Automating Insights via GenAI, RAG & Predictive Analytics, and also building AI-powered Chatbots. With experience in both multinational corporations and startups, I understand the challenges newcomers face in generative AI.