Ruchir Puri is chief scientist of IBM Research, an IBM Fellow and a vice president for IBM’s technology and technical community. His work spans automated chip design and AI systems for understanding and generating software.[1]
Start with Project CodeNet →
Education
Bachelor’s degree, Electronics and Communication Engineering
Completed 1988
[3]
Links
IBM Research LinkedInChief scientist, IBM Research; IBM Fellow; vice president, technology and technical community · IBM
Career
Puri’s early work concerned methods for designing high-performance electronic circuits and microprocessors. His institutional biographies describe later leadership across IBM’s research and AI businesses, including serving as Watson’s chief technology officer and chief architect from 2016 to 2019. He earned his master’s degree at IIT Kanpur and doctorate at the University of Calgary.[1][2][3]
Contributions
Puri is the first listed author of CodeNet, a dataset of programming-problem submissions in many languages. It pairs code with annotations and, for most samples, input-output tests. Researchers can use this material to compare programs, translate between languages or study execution costs. The test cases offer a way to check behavior on supplied inputs; they do not prove correctness for every possible input.[4]
Ideas
Puri’s career connects two forms of automation: tools that help design computer hardware and models that help work with software. CodeNet makes part of that second task concrete by preserving unsuccessful submissions and execution information alongside working programs, supporting research on errors as well as successful examples.[4][1]
Tools, writing & talks
Tools
Start here
Project CodeNet
github.comA multilingual code dataset with problem metadata and test cases for AI-for-code research.
Begin with the dataset organization and one programming problem’s submissions and tests. Basic programming is enough to inspect examples; model training requires a separate experimental setup.