Learning Tracks

Math Essentials

The core math behind deep learning — activation functions, softmax, loss functions, entropy, KL divergence, and probability distributions — explained visually with code and interactive plots.

PyTorch Cheatsheet

From tensors to transistors — memory layout, autograd, the compilation stack, torch.compile, and how your Python code becomes GPU microcode.

GPU Architecture & CUDA

From transistors to kernels — streaming multiprocessors, the CUDA programming model, the software stack, and the roofline model for performance analysis.

Transformers

From the attention mechanism to full encoder and decoder architectures — how transformers process sequences, why each component exists, and how to build one from scratch.

NanoGPT Speedrun

Incremental improvements that push GPT pre-training efficiency to its limits — from baseline to SOTA in hours.

Fine-tuning

From full fine-tuning to LoRA and QLoRA — how to adapt foundation models to your task, build instruction datasets, run distributed training, evaluate results, and merge models.

RLHF & Alignment

From SFT's rigidity to reinforcement learning — how PPO, DPO, and GRPO align language models with human preferences, with formulas, implementations, and the HuggingFace TRL ecosystem.

RAG Pipelines

From TF-IDF and BM25 to dense bi-encoders, hybrid fusion, rerankers, HNSW indexing, and fine-tuning with contrastive losses — everything you need to build and understand production-grade retrieval-augmented generation systems.

Inference Optimization

KV-cache, speculative decoding, quantization (GPTQ, AWQ, GGUF), continuous batching, and the engineering behind serving LLMs at scale with low latency.

Vision-Language Models

From CLIP to LLaVA — contrastive pre-training, Vision Transformers, SigLIP, DINOv2, multimodal fusion, and visual instruction tuning.

Vision-Language-Action Models

VLAs for robotics — grounding language and vision into motor policies, from OpenVLA to diffusion-based action prediction.

Audio & Omni Models

From waveforms and Fourier transforms to neural audio codecs, speech synthesis, and omni-modal models — how machines hear, speak, and reason about sound.

Image & Video Generation

From DDPM to latent diffusion, flow matching, and Diffusion Transformers — how Stable Diffusion, Flux, DALL-E, and Sora generate images and video from text.

Long Context & Memory

From the quadratic attention wall to RoPE scaling, sparse attention, and memory-augmented transformers — how models handle long sequences and remember beyond the context window.

Agents & Tool Use

From function calling and ReAct to MCP, computer use, and multi-agent orchestration — how LLMs go from generating text to taking actions in the real world.