Prithviraj Singh Shahani

Profile photo

Hello! I'm Prithvi, a Software Engineer at Neon Health.

I care about how we can meaningfully evaluate and improve the safety of AI systems at scale.

I graduated from Tufts University with a degree in Computer Science and Cognitive Science. While I was there, I founded and directed Tufts AI Safety.

Outside of work, I enjoy Brazilian jiu-jitsu, ultimate frisbee, reading, taking care of my plants, and exploring San Francisco. Feel free to reach out!

Papers

Noise Injection Systemically Degrades Large Language Model Safety Guardrails
Prithviraj Singh Shahani, Kaveh Eskandari Miandoab, Matthias Scheutz
arXiv
Noise Injection Reveals Hidden Capabilities of Sandbagging Language Models
Cameron Tice, Philipp Alexander Kreer, Nathan Helm-Burger, Prithviraj Singh Shahani, Fedor Ryzhenkov, Fabien Roger, Clement Neo, Jacob Haimes, Felix Hofstätter, Teun van der Weij
NeurIPS, 2025
arXiv
Probing a vision-language-action model for symbolic states and integration into a cognitive architecture
Hong Lu, Hengxu Li, Prithviraj Singh Shahani, Stephanie Herbers, Matthias Scheutz
IEEE International Conference on AI and Data Analytics (ICAD), 2025
arXiv

Projects

Projects coming soon!