Foundations of Generative AI
-
Introduction, Key Concepts, and Evolution
-
Applications in Healthcare, Finance, Media, Education, Entertainment
-
Types of Generative Models: VAEs, GANs, Diffusion Models, LLMs
-
Tools & Frameworks: PyTorch, TensorFlow, Hugging Face, LangChain
-
Hands-On: Setting up development environment
Generative Adversarial Networks (GANs)
-
Fundamentals: Generator & Discriminator
-
Loss Functions, Training Dynamics
-
Variants: DCGAN, CycleGAN, StyleGAN
-
Applications: Deepfakes, Art, Style Transfer, Image-to-Image Translation
-
Hands-On: Build a GAN; Experiment with StyleGAN
Variational Autoencoders (VAEs)
-
Encoder-Decoder Architecture & Latent Space
-
Applications: Anomaly Detection, Data Compression, Image Generation
-
Hands-On: Create a VAE; Explore multi-modal VAEs (text-to-image)
Large Language Models (LLMs) & NLP
-
Transformer Architecture & Attention
-
Pre-training vs. Fine-tuning
-
Tokenization, Embeddings
-
Applications: Chatbots, Summarization, Translation, Code Generation
-
LangChain for Prompting, Chaining, and Agent Workflows
-
Retrieval-Augmented Generation (RAG) with Vector DBs (Pinecone, Weaviate)
-
Hands-On: Build a GPT-like model; Fine-tune LLMs; RAG-based Chatbot
Diffusion Models
-
Theory: Forward & Reverse Diffusion
-
Tools: Stable Diffusion, DALL-E, ControlNet
-
Applications: Text-to-Image, Image-to-Image, Video Generation (Sora)
-
Hands-On: Image/Video generation with Stable Diffusion & ControlNet
Image, Video & Multi-Modal AI
-
Photorealistic Image Generation
-
Video Models for Gaming/Entertainment
-
Multi-modal AI: Combining Text, Image, Audio (CLIP, ALIGN)
-
Hands-On: Image generator, AI video creation, multi-modal captioning
Training & Optimization
-
Training Techniques: Regularization, Dropout, Early Stopping
-
Hyperparameter Tuning: Grid, Random, Bayesian Optimization
-
Fine-Tuning Pretrained Models: Transfer Learning, PEFT (LoRA, QLoRA)
-
Evaluation Metrics: FID, Inception Score, BLEU, ROUGE
-
Hands-On: Fine-tune open-source models; PEFT on custom dataset
Deployment & AI Agents
-
Web & Mobile Integration
-
Serverless Deployment (AWS Lambda, GCP Functions)
-
API Development (Flask, FastAPI, Streamlit, Gradio)
-
LangChain for AI Agents & Tool Integration
-
Monitoring: Drift, Feedback Loops, Retraining
-
Hands-On: Deploy a GenAI app; Build LangChain-powered agent
Ethics & Limitations
-
Deepfakes, Copyright, and Bias in AI
-
Security, Privacy, and Regulations (GDPR, CCPA)
-
Intellectual Property Issues
-
Guest Lecture: AI Ethics & Policy Expert
Capstone Projects
-
Build a full Generative AI application
-
Example Projects:
-
RAG-powered Chatbot
-
Marketing Content Generator
-
Multi-modal AI System (text + image + audio)
-
-
Hands-On: End-to-end project, documentation, and presentation
Course Outcomes:
-
Gain strong knowledge of GANs, VAEs, Diffusion Models, and LLMs.
-
Build, fine-tune, and deploy generative models for text, image, video, and multi-modal tasks.
-
Use tools like PyTorch, TensorFlow, Hugging Face, and LangChain for real-world AI solutions.
-
Apply ethical, legal, and responsible AI practices.
-
Create industry-ready projects for roles like AI Engineer and LLM Developer.
