Ameya Velingker (अमेय वेलिंगकर)
Research Scientist
I am a research scientist whose research interests are broadly in the area of machine learning and theoretical computer science. I am currently interested in reasoning in ML models (e.g., AI for mathematics/science) and machine learning on graph-structured and relational data, combining tools from algorithms with machine learning. Other areas I have worked in include streaming algorithms, privacy, error-correcting codes, etc. I was previously a research scientist at Google Research (until 2024), where my work has been used in a number of systems such as AlphaProof (theorem proving), Google Maps (routing/navigation), and Gboard (private analytics).
I received my PhD in Computer Science in 2016 at Carnegie Mellon University, where I was advised by Venkatesan Guruswami and Gary Miller. Afterwards, I was a Research Scientist at École Polytechnique Fédérale de Lausanne (EPFL) from 2016-2018.
In 2011, I completed the Master of Advanced Study in Mathematics at the University of Cambridge (Trinity College) under the support of the Gates Cambridge Scholarship. Prior to that, I received an AB in Mathematics and SM in Computer Science (supported by a Siebel Scholars Award) from Harvard University in 2010.
Some articles highlighting my work: AlphaProof (NYT, MIT Technology Review, Ars Technica, The Hindu, The Guardian, GDM Blog), SecAggIBLT (Google Research Blog), Exphormer (Google Research Blog), Maps (1, 2).
Other links: My ICML 2024 tutorial on Graph Learning: Principles, Challenges, and Open Directions (with Adrián Arnaiz-Rodríguez)
- Machine learning on graphs
- Reasoning in AI models (e.g., AI for mathematics/science, theorem proving)
- Algorithms
News
- Our paper, Even Sparser Graph Transformers, has been accepted to NeurIPS 2024.
- The full video for our ICML 2024 tutorial on Graph Learning: Principles, Challenges, and Open Directions is now available on YouTube!
- We recently announced our work on AI models AlphaProof and AlphaGeometry 2, which managed to achieve a silver medal level of performance at the 2024 International Mathematical Olympiad (IMO)!
- I presented a tutorial, Graph Learning: Principles, Challenges, and Open Directions, at ICML 2024 along with Adrián Arnaiz-Rodríguez. Details can be found here.
- Our paper, Weisfeiler-Leman at the margin: When more expressivity matters, has been accepted to ICML 2024.
- Our work on graph transformers (Exphormer) appeared on the Google Research blog!
- I was invited to give a talk on at the 2024 workshop on Algorithmic Aspects of Neural Networks in Cologne, Germany, where I spoke about Oversquashing in Graph Neural Networks.
- Our paper, Locality-Aware Graph Rewiring in GNNs, has been accepted to ICLR 2024.