Tengyu Ma
Tengyu Ma is an assistant professor of computer science at Stanford and Chief AI Scientist at MongoDB whose work spans machine learning theory, reinforcement learning and automated theorem proving. He earned a computer science bachelor's degree at Tsinghua in 2012 and completed a Princeton doctorate under Sanjeev Arora by 2018. His work on matrix completion received a NIPS student paper award in 2016, and his dissertation received a 2018 ACM honorable mention. In 2023, he co-founded Voyage AI to build retrieval systems for AI applications; MongoDB acquired the company in February 2025. With Kefan Dong, he coauthored STP in 2025, a theorem prover that learns by proposing conjectures and attempting proofs.
- Now at
- Stanford University and MongoDB
- Works
- 160
- Citations
- 11,758
- h-index
- 43
Publication numbers from OpenAlex, as of 28 October 2026.
Timeline
-
2012
Computer science at Tsinghua
He earns a BS in computer science at Tsinghua University in 2012, graduating from the Yao Class.
-
2016
Matrix completion and optimization
In 2016, he coauthors Matrix Completion has No Spurious Local Minimum with Rong Ge and Jason Lee. The paper proves that all local minima of a commonly used matrix completion objective are global minima and receives the NIPS 2016 Best Student Paper Award.
-
2018
Assistant professor at Stanford
By 2018, he completes a PhD in computer science at Princeton University under Sanjeev Arora. A September 2018 Stanford report announces his appointment as an assistant professor of Statistics with a joint appointment in Computer Science.
-
2018
An ACM dissertation honour
His dissertation, Non-convex Optimization for Machine Learning: Design, Analysis, and Understanding, receives an honorable mention for the 2018 ACM Doctoral Dissertation Award.
-
2020
Offline reinforcement learning with MOPO
In 2020, he coauthors MOPO: Model-based Offline Policy Optimization. The method learns policies from previously collected data and penalizes rewards using uncertainty in the learned dynamics.
-
2021
A Sloan Research Fellowship
He receives a Sloan Research Fellowship in 2021. His research includes deep learning theory, reinforcement learning, representation learning and non-convex optimization.
-
2023
Co-founding Voyage AI
In 2023, he co-founds Voyage AI to help developers build retrieval systems for AI applications.
-
February 2025
Voyage AI joins MongoDB
MongoDB acquires Voyage AI in February 2025 to integrate its embedding and reranking models with MongoDB's database.
-
2025
The Self-play Theorem Prover
With Kefan Dong, he coauthors STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving at ICML in 2025. The system trains a conjecturer and a prover that supply training signals to each other, and the team evaluates it with Lean and Isabelle.
-
As of October 2026
Stanford professor and MongoDB scientist
As of October 2026, he is an assistant professor of computer science at Stanford University and Chief AI Scientist at MongoDB.
Career
- NowMongoDBChief AI Scientist
- NowStanford UniversityAssistant Professor of Computer Science
- From 2023Voyage AICo-founder
- 2017–2019Meta
- Stanford UniversityAssistant Professor of Statistics with a joint appointment in Computer Science
Education
- 2012Tsinghua UniversityBS in Computer Science
- Princeton UniversityPhD in Computer Science
Sources and credits
Photo: We have no published photo of Tengyu Ma yet. Send us a free-licensed photo.
- Computer Science Celebrates Siebel Scholars | CS
- IIIS Alumnus Awarded a 2021 Sloan Research Fellowship-Institute for Interdisciplinary Information Sciences (IIIS)
- Matrix Completion has No Spurious Local Minimum
- Awards
- Tengyu Ma appointed Assistant Professor of Statistics | Department of Statistics
- Tengyu Ma's Profile | Stanford Profiles
- Tengyu Ma receives honorable mention for ACM Doctoral Dissertation Award | Department of Statistics
- [2005.13239] MOPO: Model-based Offline Policy Optimization
- Stronger Together: Why We Chose to Join MongoDB – Voyage AI
- MongoDB Announces Acquisition Of Voyage AI To Enable Organizations To Build Trustworthy AI Applications | MongoDB
- STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving
- Tengyu Ma | Computer Science
- What Is Agent Memory? A Guide to Enhancing AI Learning and Recall | MongoDB

