Sharan Maiya

I’m a first year PhD student in the Language Technology Lab at the University of Cambridge, where I work on technical alignment research and evaluations for large language models. I’m supervised by Anna Korhonen.

I’m also interested in climate change mitigation, and how AI can accelerate research and decision-making in climate science. I’m affiliated with CamCID and climaTRACES through my co-supervisor Ramit Debnath and the AI4ER CDT.

Currently I’m working on benchmarking and building truthfulness-detection methods for large language models as a co-founding research engineer at Cadenza Labs. I’ll also be working on evaluations with Evan Hubinger as an incoming Winter 2024-25 MATS scholar. In a previous life I worked on causal discovery algorithms and targeted causal learning, looking at the health effects of exposure to air pollution.

A lot of my time is spent thinking about AI alignment, whatever movie I watched yesterday, and how much I miss Scottish tap water, so reach out if you want to talk about any of those (or perhaps about research, I guess). You can do that at firstname98lastname[at]gmail.com or through the links below.