Bio

Hello, I’m an engineer with a background in applied mathematics and computer science, and a fondness for software engineering. My current work focuses on artificial intelligence, particularly in enhancing mathematical reasoning and problem-solving.

During my Master’s thesis at JP Morgan, I worked on applications of artificial intelligence to mathematical reasoning tasks. I developed a custom symbolic algebra system used for generating synthetic data in natural language, which was part of our research on leveraging large language models for mathematical reasoning. Additionally, I worked on a meta-programming framework in Lean for generating synthetic data using forward reasoning.

I am now pursuing this work as a full-time AI researcher at JP Morgan. I am interested in tackling mathematical reasoning tasks from both informal and formal ends. On the one hand by improving the capabilities of large language models through training, tool use, and data curation and generation. On the other hand, by using interactive theorem provers to formalize and verify mathematical reasoning tasks. I find the formal approach particularly interesting because in principle it doesn’t mandate to define any specific mathematical objects to be manipulated, i.e. it’s a field where the model can self-play (define objects, prove theorems on them, all verified by the theorem prover without human intervention).

I am also interested in the more technical aspects of AI, such as distributed computing and kernel optimization.

Feel free to explore this website where I try to share a bit of what I do, in terms of academic work, research, open-source projects, and other interests. It can also be seen as an extension of my CV.