From foundational LLM concepts to building production-grade AI systems. Master transformers, RAG, agents, diffusion models, and more.
A structured journey through four progressive phases
16 weeks of structured, progressive learning
Understand what Large Language Models are, their history, capabilities, and the landscape of modern AI. Explore key concepts like pretraining, prompting, and emergent abilities.
Discover how text becomes numbers. Learn tokenization algorithms (BPE, WordPiece), word embeddings, vector spaces, and the revolutionary self-attention mechanism.
Deep dive into the Transformer architecture. Explore multi-head attention, feed-forward layers, layer normalization, positional encoding, and how all components fit together.
Understand causal (masked) attention for autoregressive generation. Build a working Transformer from scratch in Python, implementing every component hands-on.
Explore how LLMs are trained on massive datasets. Learn about pretraining objectives, distributed training, data pipelines, compute requirements, and scaling laws.
Make models smaller and adapt them to your needs. Learn quantization techniques (INT8, INT4, GPTQ, GGUF), LoRA, QLoRA, and practical fine-tuning workflows.
Learn the theory behind RAG. Understand vector databases, embedding models, chunking strategies, retrieval methods, and how to ground LLM responses in real data.
Build a complete RAG system from scratch. Implement document processing, vector storage with ChromaDB/Pinecone, retrieval pipelines, and evaluate RAG quality.
Build autonomous AI agents that can use tools, browse the web, execute code, and interact with APIs. Understand ReAct patterns, function calling, and agent architectures.
Explore the Model Context Protocol, advanced context window management, and orchestrating multiple specialized agents working together on complex tasks.
Ship AI to production. Learn evaluation frameworks, monitoring, guardrails, cost optimization, latency management, and best practices for reliable AI applications.
Explore models that think step-by-step. Understand chain-of-thought reasoning, tree-of-thought, self-consistency, and how models like o1 and DeepSeek-R1 achieve deeper reasoning.
Venture beyond text. Explore vision transformers (ViT), CLIP, multimodal models, image generation architectures, and video understanding systems.
Master the math and mechanics of diffusion models. Understand denoising, U-Net architectures, latent diffusion, Stable Diffusion, and controllable generation.
Put it all together. Design, build, and deploy a complete AI engineering project that demonstrates mastery of concepts from all previous weeks.
Consolidate your knowledge with overarching principles. Cover ethics, safety, responsible AI, career paths, staying current, and the future of AI engineering.
What you should know before starting
Comfortable with Python basics: functions, classes, data structures, and common libraries like NumPy.
Linear algebra fundamentals (vectors, matrices), basic probability, and calculus concepts.
Basic terminal/command line usage. Ability to install packages and run scripts.
Familiarity with machine learning concepts is helpful but not required. We cover fundamentals as needed.