Mini Courses
Hands-on, step-by-step mini-courses: each lesson is a single focused session — intuition-first explanation, a deep code walkthrough with the methodology explained line by line, theme-matched visuals, and a 🧪 Your task exercise with a hidden solution. Start with the encyclopedia for theory, then pick a course and build.
📚 The AI & ML Encyclopedia
The flagship reference — 47 chapters across 14 parts. The whole field in one continuous, intuition-first guide: math foundations → classical ML → deep learning → LLMs → agents → MLOps, with interactive demos, animated visuals, cheat sheets, and cross-linked chapters. The theory companion to every course below.
🔥 Deep Learning with PyTorch
9-Lesson mini-course. From raw tensors and autograd to a trained, exported, served model — the canonical PyTorch 2.x path, built by hand.
⚡ Building Transformers from Scratch with PyTorch
10-Lesson mini-course. Hand-write every component of a GPT-style decoder — tokenizer, attention, blocks, training, KV-cache generation — and train your own tiny GPT.
📊 Deep Learning with TensorFlow & Keras
9-Lesson mini-course. TensorFlow 2.x / Keras 3 end to end — tensors, tf.data pipelines, training loops, CNNs, transfer learning, and deployment to Serving/TFLite.
🚢 ML in Production — MLOps
10-Lesson mini-course. One model, taken from notebook to production: reproducible training, MLflow tracking & registry, Docker, FastAPI, vLLM, CI/CD gates, drift monitoring, continuous retraining.
📖 Build Your Own Wikipedia LLM
11-Lesson mini-course. From a raw Wikipedia dump to your own instruction-tuned model — cleaning pipeline, from-scratch PyTorch pretraining with W&B monitoring, synthetic SFT data on GitHub, DPO alignment, and serving — end to end on rented vast.ai GPUs (~$15–30 total).