Generative AI & Large Language Models
From Fundamentals to Real-World Model Development
Duration
3Days
Level
Advanced Level
Design and Tailor this course
As per your team needs
Overview
This immersive, hands-on program provides a comprehensive introduction to Generative AI and Large Language Models (LLMs). Designed for modern technology teams, the course blends theoretical foundations with practical implementation to help participants confidently design, build, and apply generative AI solutions.
Participants will explore how state-of-the-art models generate text, images, and other outputs, understand the architectures that power them, and gain hands-on experience using industry-relevant tools and frameworks.
Audience
- Software Developers
- Data Engineers & Data Analysts
- AI/ML Engineers
- Technical Product Owners
- Solution Architects exploring GenAI adoption
Prerequisites
Participants should have:
- Strong working knowledge of Python
- Familiarity with NumPy, Pandas, and scikit-learn
- Understanding of data analysis and machine learning fundamentals
- Basic exposure to deep learning concepts (recommended)
Curriculum
- What is Generative AI and where it is used
- Types of generative models and real-world applications
- Overview of GANs, VAEs, and autoregressive models
- Neural networks and architectures
- Optimization techniques (gradient descent, backpropagation)
- Key deep learning concepts required for GenAI
- Probability distributions and sampling
- Latent spaces and representation learning
- Hands-on: Implementing a simple generative model using Python
- VAE architecture and intuition
- Training and generating new samples
- Hands-on: Image generation and reconstruction using VAEs
- GAN theory and training dynamics
- Generator vs Discriminator
- Synthetic data generation
- Hands-on: Training and evaluating a GAN
- How modern LLMs work
- Transformers as the foundation of LLMs
- Popular transformer-based models
- Prompt Engineering fundamentals
- Hands-on: Text summarization and grammar correction using prompts
- Limitations of early Seq2Seq models
- Attention mechanisms and intuition
- Scaled dot-product attention
- Implementing attention in deep learning frameworks
- Hands-on: Seq2Seq translation with attention
- Why transformers replaced traditional Seq2Seq
- Multi-head attention
- Transformer architecture explained
- Hands-on: Building a transformer model from scratch
- Introduction to open-source model hubs
- Using pre-trained models and datasets
- Uploading and managing model checkpoints
- When to train vs reuse existing models
- Compute considerations and cost trade-offs
- Full fine-tuning and catastrophic forgetting
- Single-task vs multi-task fine-tuning
- Transfer learning strategies
- Hands-on:
- Fine-tuning a language model and validating results
- Transfer learning for sentiment analysis
Duration
3Days
Level
Advanced Level
Design and Tailor this course
As per your team needs
