A learning platform for people doing research in artificial intelligence.
Rigorous, code-forward courses on transformers, diffusion, RLHF, and ML systems. Taught by researchers who publish in the venues you read.
Derived, not hand-waved
Every technique starts from the math on paper before touching code.
Runs on your hardware
Notebooks target a single 24GB GPU. Nothing requires a cluster.
Papers as the syllabus
Each module ships with the arXiv IDs it reimplements.
Reproducible labs
Docker + seeded environments so your run matches the video.
Start here.
Transformers From Scratch
Build a decoder-only LLM in PyTorch, one matrix multiply at a time.
RLHF & Modern Alignment
PPO, DPO, and the mechanics of instruction-tuning frontier models.
Diffusion Models In Depth
Score matching, DDPM, DDIM, and the math behind Stable Diffusion.
"I've paid for a lot of ML courses. This is the first one that made me feel like I was in a research group instead of a bootcamp."
Enroll in cohort 026.
One-time payment. Lifetime access. New cohorts every two months with live office hours.