Madhav Prashanth Ramachandran
I stare at numbers and figure out what they mean.
Twelve years in AI/ML. The deeper I go, the more I realize I know nothing. So I work through it publicly — one topic, one derivation at a time.
The power of seeing resonates with me more than the power of words. Language describes the world. For me, images are the world.
Writing
- Transformers & Perception — Self-attention from covariance geometry to ViT, DINO, DETR, and beyond.
- 3D Computer Vision — Classical and neural 3D computer vision.
- World Models — Representation learning, VICReg, JEPA.
- Survival Analysis — Hazard functions, Kaplan-Meier, Cox-PH.