Manooshree Patel

University of California – Berkeley | Science and Mathematics Education (SESAME)

Contact: manooshreepatel (at) berkeley (dot) edu

I am interested in combining advances in AI4Math and AI4Education to develop classroom solutions that are: 1) scalable to large classrooms (1000+ students) 2) increase student access to high quality engineering & math education.

Currently, I am developing LeanTutor. LeanTutor is an end-to-end Lean4-based tutoring system that formally verifies student-written math proofs, generates correct next steps, and provides the appropriate instructional guidance. LeanTutor is composed of three modules: (i) an autoformalizer/proof-checker, (ii) a next step generator, and (iii) a natural language feedback generator. This system combines multiple of my research interests including developing deep learning methods for theorem proving (autformalization, proof generation), defining and reproducing a “good tutor” in an undergraduate math classroom, and exploring the interaction and interface between a user and a “math-machine”. pre-print and source code to be released

My past research projects in education champion accessibility. I have worked on a project understanding the impact of an inclusive study group formation system deployed to thousands of engineering undergraduates, a project developing a tactile tool for BVI (Blind and Visually Impaired) to learn fundamental algebraic manipulations, and a project which evaluated the impact of course-based research mini-projects on students self-beliefs about conducting engineering math research.

Publications:

Patel, M., Bhattacharyya, R., Lu, T., Mehta, A., Voss, N., Norouzi, N., & Ranade, G. (2025). LeanTutor: A formally-verified AI tutor for mathematical proofs. Under Review at COLM 2025.

Agyare, B., Patel, M., Matsumoto, A., & Ranade, G. (alphabetical) (2025, February). Broadening Participation in CS Research with Scalable Undergraduate Research Mini-Projects. In Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 2 (pp. 1355-1356).

van Leendert, A., Patel, M., Lambert, S. G., Abrahamson, D., Gorlewicz, J. L., & Shvarts, A. (2024). Embodied learning of algebra for Braille readers. In K. Beswick (Ed.), Proceedings of the 15th International Congress on Mathematical Education.

Agyare, B., Matsumoto, A., Patel, M., & Ranade, G. (alphabetical) (2023). Student feedback on opt-in, inclusive, course-integrated study groups. IEEE: Frontiers in Education.

Patel, M., & Chalageri, S. (2023). GlotBot: Hybrid language translator for secondary level mathematics classrooms. In Proceedings of the 22nd Annual ACM IDC.