Welcome to my space for exploring ideas across AI, aviation, geopolitics, philosophy, and culture. Each blog represents a different facet of my interests and work.
A neural system that learns to parse mathematical notation rather than compute it, achieving 90%+ accuracy by separating structure understanding from calculation.
When you train a neural network, you're running a dynamical system that carves out a representation space. ToDACoMM measures the topology of what gets carved, revealing a striking divide between encoders and decoders, and now extends to MLPs and large-scale transformer analysis.
A comprehensive look back at my year in AI - scaling up at work, building the Praval agentic AI framework, Vajra Search, Tlon mathematics, ToDACoMM, exploring mechanistic interpretability, honest research failures, and reflections on curiosity vs. accomplishment.
Applying perturbation analysis and Lyapunov exponents to neural network training. Dense networks converge; transformers diverge. The architecture determines the stability.
After months of development, Vajra BM25 achieves ~1.2-1.3x faster latency than BM25S while maintaining competitive accuracy. I share what I learned building it and benchmark results across BEIR and Wikipedia datasets.
Building a mathematical framework where processes are primitive and objects emerge as stable patterns - with 20 axioms, rigorous proofs, and simulations of dynamical systems that demonstrate the core insight: stability is special, not generic.