I'm an AI safety researcher working in Cambridge, MA.

My research focuses on the physics of many-body systems. Currently, I'm interested in biological and artificial neural networks.

Broadly, I'm interested in how the universe computes.

I'm a senior studying physics at the University of Illinois, Urbana-Champaign.

Some notes on various topics. My blog with longer thoughts.

research

• Decoder-Preserving Sparse Autoencoders: Which Readouts Survive Sparse Compression?
  A. Deshpande
  New England Mechanistic Interpretability Workshop 2026. Paper Code
• Fermionic Shadow Regression: Amortized Learning of Molecular Electron Dynamics
  A. Deshpande, K. Panicker, L. Mantilla Calderón, M. Bagherimehrab, A. Aspuru-Guzik
  Talk at Quantum Machine Learning at IEEE Quantum Week 2026 .
• Feature-Resolved Attention
  D. Manning-Coe, N. Aljaafari, J. Stephenson, K. Kapre, H. Jain, A. Deshpande
  ICML Mechanistic Interpretability 2026.
• Crosscoding Through Time: Sparse Feature Discovery Across Sequence Positions
  D. Manning-Coe, H. Xuanyuan, A. Deshpande, A. Shportko, W. Fei
  NeurIPS 2026. Paper Code