Manya Bansal

I’m part of the founding team at humans&, where I lead the user modeling team with Alexis Ross.
Before humans&, I was a Ph.D. student at MIT, where I worked with Saman Amarasinghe and Jonathan Ragan-Kelley on programming languages for high performance computing. My research focused on helping programmers write fast, low-level GPU code that is still modular, reusable, and correct. I believe programming systems should give people precise control over hardware while making that control easier to compose and trust.
I received my B.S. in pure mathematics from Stanford in 2023. There, I was lucky to work with Fredrik Kjolstad and Dawson Engler, who let me tinker and build many, many small things.
News
- Oct 2026 Built and launched &Tutor, a model trained to teach.
- Sep 2026 Launched Persimmon, a model that simulates how people behave.
- May 2026 Our paper “Modular GPU Programming with Typed Perspectives” received a Distinguished Paper Award at PLDI'26.
- Jan 2026 Joined humans& and went on leave from my Ph.D. program at MIT.
- Dec 2025 Wrapped up TA-ing the second offering of MIT’s GPU programming course.
- Nov 2025 Awarded the NVIDIA Graduate Fellowship (declined).