Scholar
Rohan Subramani
Google Scholar ID: O-qEYjUAAAAJ
University of Toronto
Citations & Impact
Citations
17
H-index
3
i10-index
0
Publications
4
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Published papers include 'Higher-Order Beliefs in Incomplete Information MAIDs', 'The Partially Observable Off-Switch Game', 'Generalization Analogies: A Testbed for Generalizing AI Oversight to Hard-To-Measure Domains', and 'On The Expressivity of Objective-Specification Formalisms in Reinforcement Learning'; projects such as coding GPT-2 from scratch, implementing basic (and not-so-basic) LLM agents, completing ARENA exercises, experimenting with neural network pruning.
Research Experience
- Founder of Aether, an independent research lab focused on foundation model agent safety; interned with AI Safety Hub Labs (now LASR Labs), UC Berkeley's Center for Human-Compatible AI (CHAI), and the ML Alignment & Theory Scholars (MATS) program.
Education
- Incoming PhD student at the University of Toronto, supervised by Professor Zhijing Jin; completed an undergrad in CS and Math at Columbia.
Background
- Aiming to promote welfare and reduce suffering for all sentient beings, focused on AGI safety research; particularly interested in foundation model agents (FMAs).
Miscellany
- Hobbies include playing tennis, listening to rock and indie pop music, playing social deduction games, reading fantasy books, watching a variety of TV shows and movies, and playing the saxophone.
Co-authors