Undergraduate building from the ground up toward world model research — the kind of work that lets models understand, generate, and reason about persistent 3D environments. Starting with diffusion internals and moving toward spatial intelligence.
- 🖼️ Currently: image diffusion — LDMs, SDEs, DiT architectures, built from scratch
- 🎬 Next: video generation — temporal consistency, spacetime patching, scene persistence
- 🧊 Then: 3D spatial generation — Gaussian Splatting, novel view synthesis, Score Distillation Sampling
- 🌍 Goal: Large World Models — generative models that build and navigate coherent 3D worlds
Core ML
Deployment & Tooling
