Suggestions
Sumit Gulwani
Computer Scientist at Microsoft
Sumit Gulwani is a distinguished computer scientist and Partner Research Manager at Microsoft, where he has been working since August 2005.1 He is based in Redmond, Washington, United States.
Professional Background
Gulwani is known for his pioneering contributions to the field of programming languages, particularly in program analysis and program synthesis. His work has focused on empowering computer users worldwide to be more productive and educated through innovative technologies.
Key Achievements
Research Contributions:: Gulwani has made significant advancements in:
- Transforming imprecise human intent into program specifications
- Developing algorithms for synthesizing string transformation programs
- Applying program synthesis techniques to intelligent tutoring systems
Notable Work::
- He developed the technical basis for "Flash Fill" in Microsoft Excel, which allows users to perform complex and repetitive programming tasks without programming knowledge.1
Awards and Honors
Gulwani has received numerous prestigious awards for his work:
- ACM Fellow (2024) for contributions to AI-assisted programming2
- Max Planck-Humboldt Medal (2021) for automatic programming and computational education1
- ACM SIGPLAN Robin Milner Young Researcher Award (2014)1
- ACM SIGPLAN Doctoral Dissertation Award (2005)1
- President's Gold Medal from IIT Kanpur (2000) for best academic performance in the graduating class1
Education
Gulwani received his education from:
Current Focus
As a Distinguished Scientist at Microsoft, Gulwani is currently working on AI and Copilots, continuing to innovate in the field of AI-assisted programming with applications to developer productivity.3
Highlights
Had a blast discussing #AI-assisted Programming at the Marktoberdorf Summer School this week. 4 lectures on 4 different (hot) days meant I didn’t need to select which of my PROSE-Team Tshirt to sport, each of which blended in seamlessly as a "human slide" in the presentation.
•The classic Octember #FlashFill shirt never fails to amuse and motivated the powerful neuro-symbolic techniques that power the FlashFill++ feature in my first lecture on programming by input-output examples. •The Code Clippy, aka “#IntelliCode Suggestions” shirt (design credits: Gustavo), highlighted the importance of temporal context in predicting user intent for software engineering tasks. •The Sisyphus shirt (creative design credits: Gust, Param) motivated the grand challenge of developing an AI bot that can guide students like a human TA in #Education settings. •The last lecture on programming-by-NL made me realize that natural language is useful not just as an input modality, but also as an output artifact (program explanations, repair hints) and, most significantly, as a communication mechanism. It aids not only in resolving ambiguity or collaborative problem-solving with the user but also in thinking, planning, and reflection (#agentic architectures). Most of this work has been led by our research fellows in India. The Word-Cloud shirt, which helped me summarize the key themes, was a tribute to them—they designed and gifted me this shirt during my last visit to Bangalore.
These lectures required me to steal time from other responsibilities, but it was all worth it. Not only did I discover new ideas in the process, but the journey also strengthened my friendships with fellow lecturers and forged new connections with the students. Feeling grateful! Thank you Alexander and Peter for the invitation and your wonderful hospitality.
Link to the slides are in the comments below.
I’m honored and excited to be recognized as an #ACM Fellow for my work on #AI-assisted programming with applications to developer productivity, end-user programming, and programming education. Congratulations to all the other awardees for their richly deserved recognition.
I am grateful to my PROSE team at #Microsoft for fostering a culture of doing better science through exploring ideas as well as building robust implementations. I am grateful to my product collaborators at Microsoft for making room for these implementations inside their products and helping evolve them through telemetry-driven iterations (including Flash-Fill and Formula-by-Example features in #Excel, Data-Connectors in #PowerQuery, IntelliCode-Suggestions feature in #VisualStudio, and recently the various CoPilots). I am grateful to Microsoft executives who believed in and sponsored my research work (thanks Henrique Malvar, Joseph Sirosh, James Phillips, Sid Jayadevan, Giovanni Della-Libera, Julia Liuson, Amanda K. Silver, Benoit Barabe, Yair Helman). I am grateful to my friends and family for their curiosity about my work, and especially to my (late) father whose “how does your research help other people?” challenge inspired me to pick “simplest problems with most impact”.
A big shout out also to all of my academic collaborators, especially Ashish Tiwari, Vu Le, Arjun Radhakrishna, Gustavo Soares, Mohammad Raza with each of whom I have had the pleasure of co-authoring 10+ research papers, and the journey is still on! I feel blessed to be living in these times when AI is advancing at an amazing pace, and I hope to help shape some part of it with my colleagues. I still feel like it's day 1.
