Duration: 4 weeks 

• Introduction to AI, deep learning, and generative models
• Differences between discriminative and generative models
• Overview of models: GANs, VAEs, Diffusion Models, Transformers
• Use cases: Text, Image, Music, and Video generation

Practical:

• Build a simple text generation model using GPT-based APIs

Learning Objectives:

By the end of this module, students will:

• Grasp essential mathematical concepts that underpin generative AI.
• Understand how tensor operations and neural network components build the foundation for generative models.
• Differentiate between key generative model architectures.
• Build and experiment with basic generative models from scratch.
• Process and prepare real-world data for use in generative tasks.
• Develop the ability to integrate APIs and explore industry-relevant datasets.

Key Industry Alignment:-

• API Integration: Prepare students for the industry with real-world API usage.
• Case Studies: Use examples from NVIDIA and OpenAI.
• Hands-on Projects: Encourage practical learning with a focus on industry readiness.

Duration: 6 weeks

• Basics of neural networks and backpropagation
• Autoencoders (AE) and Variational Autoencoders (VAE)
• Use cases in dimensionality reduction and image generation
• Exploring latent space

Practical:

• Build a VAE for digit generation using the MNIST dataset

Learning Objectives:

By the end of this module, students will:

• Understand the fundamental components of neural networks and their role in generative models.
• Build and train deep neural networks for simple and advanced tasks.
• Learn and implement Autoencoders (AEs) and Variational Autoencoders (VAEs).
• Explore latent space representations and their practical uses.
• Use neural networks and AEs/VAEs for data compression, anomaly detection, and feature extraction.

Key Industry Alignment Elements:-

• Hands-on Mini-Project: Focus on anomaly detection, a critical real-world use case.
• Industry Guest Lecture: Gain insights into real-world challenges with AEs and VAEs.
• Advanced Techniques: Exposure to beta-VAEs and hybrid models for deeper learning.

Duration: 6 weeks

• Introduction to GANs and the adversarial process
• Deep Convolutional GAN (DCGAN) and CycleGAN
• StyleGAN for high-quality image synthesis
• Challenges in training GANs and techniques for stabilization

Practical:

• Train a DCGAN to generate synthetic images of handwritten digits

Learning Objectives:

By the end of this module, students will:

• Understand the theory and working principles of GANs.
• Learn and implement DCGAN, CycleGAN, and StyleGAN models.
• Develop skills to train, stabilize, and fine-tune GANs.
• Explore real-world applications of GANs in image synthesis, simulation, and content generation.
• Gain experience with stabilization techniques to address mode collapse.

Key Industry Alignment Elements:-

• Mini-Project Focus: Hands-on development of CycleGAN for virtual e-commerce.
• Industry Guest Lecture: Insights from professionals using GANs in media and gaming.
• Advanced Techniques: Students experiment with WGAN, CycleGAN, and StyleGAN for practical exposure.

Duration: 6 weeks

• Introduction to Transformers: Attention mechanism
• Pre-trained models: GPT, BERT, and LLaMA
• Applications in text generation, summarization, and translation
• Fine-tuning language models for domain-specific tasks

Practical:

• Fine-tune a GPT model to create a chatbot for a specific industry

Learning Objectives:

By the end of this module, students will:

• Understand the architecture and working principles of transformers and language models.
• Learn to fine-tune pre-trained models like GPT, BERT, and T5 for specific tasks.
• Implement and deploy transformer-based solutions for NLP applications.
• Gain experience in building chatbots, summarization tools, and translation systems.
• Explore the challenges of bias and ethics in language models.

Key Industry Alignment Elements:-

• Hands-on Mini-Project: Build a chatbot for real-world customer support.
• Industry Guest Lecture: Gain insights from professionals using transformers at scale.
• Deployment Focus: Learn to deploy NLP tools on Hugging Face Spaces and AWS.

Duration: 4 weeks

• Introduction to diffusion models for image synthesis
• Applications in AI-generated art and image editing
• Integration with creative workflows: Animation, Logo Design, and Art Generation

Practical:

• Use Stable Diffusion models to generate art based on prompts

Learning Objectives:

By the end of this module, students will:

• Understand the theory and implementation of diffusion models
• Gain hands-on experience with Stable Diffusion and other architectures for content generation
• Explore applications of diffusion models in art, animation, and design
• Fine-tune diffusion models to create domain-specific content
• Learn advanced techniques like NeRFs, multimodal generation, and AR/VR integration

Key Industry Alignment Elements:-

• Cutting-Edge Techniques: Students gain hands-on experience with NeRFs, multimodal models, and AR/VR integration
• Real-World Applications: Build practical tools aligned with trends in art, fashion, and immersive environments
• Industry Relevance: Prepare for roles in 3D content generation, AR/VR development, and media production

Duration: 4 weeks

• Video synthesis and audio generation models
• Techniques for music composition using RNNs and Transformers
• Voice cloning and speech synthesis

Practical:

• Build a music generator using Transformers trained on MIDI datasets

Learning Objectives:

By the end of this module, students will:

• Understand the key principles of generative models for video, audio, and music creation
• Build and fine-tune models like Vid2Vid, WaveNet, and music composition transformers
• Develop skills in generating dynamic content for games, media production, and virtual performances
• Gain hands-on experience in deploying generative models in cloud environments
• Learn optimization techniques for real-time media generation

Key Industry Alignment Elements:-

• Hands-on Mini-Project: Develop a dynamic soundtrack generator for real-time applications in games or media
• Industry Guest Lecture: Gain insights from experts using AI for interactive media production.
• Real-Time Focus: Learn to optimize and deploy models for real-time generation and performance

Duration: 5 weeks

• Emerging models: Diffusion models, NeRF (Neural Radiance Fields)
• Hybrid models combining GANs with VAEs
• Reinforcement Learning with Generative Models
• Use of Generative Models in Simulations and 3D Rendering

Practical:

• Implement a NeRF-based 3D rendering project or explore hybrid GAN-VAE architecture

Learning Objectives:

By the end of this module, students will:

• Master advanced hybrid architectures such as VAE-GANs and NeRFs.
• Explore Reinforcement Learning (RL) for generative AI tasks and content control.
• Understand generative flows and normalizing flows for continuous data generation.
• Develop and deploy custom hybrid models for real-world applications.
• Gain practical experience in using generative models for autonomous systems and virtual environments.

Key Industry Alignment Elements:-

• Hands-on Mini-Project: Develop a NeRF-based 3D content generator for AR/VR environments
• Industry Guest Lecture: Gain insights from experts using NeRFs and RL in practical applications
• Real-World Applications: Prepare students for roles in 3D content creation, interactive gaming, and autonomous systems

Duration: 2 weeks

• Ethical challenges: Bias in generative models
• Deep Fakes and security risks of generative AI
• Guidelines for responsible AI practices

Practical:

• Analyze biases in generated text and propose mitigation strategies

Learning Objectives:

By the end of this module, students will:

• Understand the key ethical concerns related to generative AI, such as bias, fairness, and accountability.
• Learn how to detect and mitigate bias in datasets and models.
• Explore regulatory frameworks and guidelines for responsible AI development.
• Study case studies of AI failures and develop strategies to address similar challenges.

Duration: 4 weeks

• Model optimization for deployment on cloud and edge
• Integrating generative models with APIs
• Cost management for large-scale models
• Legal, copyright, and intellectual property issues

Practical:

• Deploy a generative AI model as an API for real-world usage

Learning Objectives:

By the end of this module, students will:

• Learn strategies for deploying generative models on cloud platforms and edge devices.
• Understand the challenges of scaling models to meet high-performance demands.
• Gain hands-on experience with Docker, Kubernetes, and Hugging Face Spaces for model deployment.
• Explore CI/CD pipelines and best practices for continuous model updates.
• Develop techniques for monitoring and optimizing deployed models.

Key Industry Alignment Elements:-

• Hands-on Mini-Project: Students deploy a generative model using modern tools and platforms.
• Industry Guest Lecture: Gain insights from engineers working on scalable AI solutions.
• Real-World Focus: Learn to build scalable, production-ready systems using Docker and Kubernetes.

Duration: 8 weeks

• Build a complete generative AI solution with commercial viability
• Work in teams or individually on a domain-specific project
• Prepare a pitch and demonstration for potential stakeholders

Practical:

• Final project submission, demo, and commercialization plan

Learning Objectives:

By the end of this module, students will:

• Apply everything they have learned to develop a commercially viable generative AI solution.
• Gain experience with end-to-end project development, including design, implementation, and deployment.
• Collaborate in teams to solve real-world business problems using generative AI.
• Learn to pitch AI products effectively to stakeholders.
• Build a portfolio project that demonstrates technical, business, and deployment skills.