Portrait of Mohammad Raza

Mohammad Raza

Researcher in AI

My interests are in the fields of AI and symbolic methods, with the goal of helping users perform complex tasks through natural interaction experiences such as providing examples, natural language instructions, multi-modal inputs, or fully automatic suggestions. My research spans a range of techniques from formal methods and machine learning, with a strong focus recently on neuro-symbolic program synthesis methods using large language models. I have led the development and productization of such techniques for task automation in a variety of domains including data science, business intelligence, office productivity, and code editing. This work has had to both practical impact in mass-market products (including Microsoft Power BI, Visual Studio Code and Excel) as well as publications in major research conferences.

Selected Publications

Overwatch: Learning Patterns in Code Edit Sequences

Yuhao Zhang, Yasharth Bajpai, Priyanshu Gupta, Ameya Ketkar, Miltos Allamanis, Titus Barik, Sumit Gulwani, Arjun Radhakrishna, Mohammad Raza, Gustavo Soares, Ashish Tiwari

OOPSLA (Object-Oriented Programming, Systems, Languages & Applications) | 2022

Interactive Program Synthesis

Vu Le, Daniel Perelman, Alex Polozov, Mohammad Raza, Abhishek Udupa, Sumit Gulwani

arXiv preprint | March 2017, Vol arXiv:1703.03539

Publications by Year