About Me

Hi, I’m Sonia — a junior at Baruch College studying Mathematics. What I care about most is figuring out the things people miss.

Specifically: the evaluation and interpretability of complex models, especially LLMs. I’m not drawn to the more glamorous work of building frontier models — I’m drawn to the quieter question of how we know whether something is actually working. What do our metrics capture, what do they miss, and what does “working well” even mean in the first place? That question doesn’t stay contained to the technical side for me — it bleeds directly into how I think about AI governance and policy, because deciding how to measure a system is already a decision about what we value in it.

That interdisciplinary instinct isn’t new. I’m not a math-olympiad-and-Stanford-CS type: my path here runs through student government, committee work, and teaching just as much as it runs through machine learning classes. The place all of this comes together most concretely is in how universities are navigating AI, which I’ve worked on directly as a design consultant for a new first-year AI literacy course at Baruch, and as a current member of the AI Leadership Initiative Core Team.

My goal is a PhD in a CS-adjacent field, working on the evaluation and interpretation of complex models.

A few unserious facts about me:

• My lucky charm is a smiley face named Bob that I draw on every exam.
• Sharks are my favorite animal — hence the GitHub profile picture.
• I love em-dashes and will never stop using them, regardless of their association with AI-generated writing.

Check out my CV to read about the type of stuff I do, and check out the GitHub to see the type of stuff I do.

If you want to chat about any of the above — evaluation, interpretability, AI policy, or just what you’re working on — find me on LinkedIn. I like listening to people talk about this stuff.