Shreshth Rajan

I study Pure Math and Philosophy at Harvard. I took leave to be the first person on the team at Forge and founding engineer at Pillar.

Research
Arithmetic Intensity Aware Quantization. Post-training quantization optimizing FLOPs/byte. 1.66x speedup. [arxiv]
Uncertainty Gated Retrieval. Test-time retrieval for semantic segmentation. +11% accuracy, 87.5% cost reduction. [arxiv]
Multiver. Multi-agent memory-augmented code verification. Beats fine-tuned models on real-world vulnerability detection. [arxiv]
Multi-Agent Code Verification. Proved why multi-agent verification works. 76% bug detection in <200 ms. [arxiv]
Social Contagion and Bank Runs. LLM agents predicting the SVB→First Republic cascade. [arxiv]
Geodesics on Finite 2-Manifolds. Discrete analog of geodesic spreading in general relativity. [paper]
Mixing Rates of Markov Chains. Spectral gap controls how fast chains forget, with an RL interpretation. [paper]
Projects
Harvy. AI academic advisor with daily active usage from over 50% of Harvard students. [site] [press]
Visdep. Codebase visualizer that turns large repos into interactive maps. [site] [writeup]
STAR. Self-learning system that cuts human labeling for code generation by 85%.
Sisyphus. Physics-integrated world models for robot RL environments.
Robomerge. First autonomous VLA preprocessing pipeline. [github]
Writing
Search Agents Don't Have a State
Codebase Understanding
Incompleteness and Intelligence
Dignāga's Basis of Cognition
India's Automation Trap

I'm an investor or advisor to Arda (seed), Etched (seed), Applied Compute (seed), Sable (seed), Transfyr AI (advisor), Advanced Machine Intelligence (seed), and Operating Intelligence (seed).

Email · GitHub · X