Timofey Bryksin (JetBrains) - Improving Code Completion by Learning from Anonymous IDE Usage Logs 

Find out about upcoming talks: bit.ly/LLMs-fo...
Title: Improving Code Completion by Learning from Anonymous IDE Usage Logs.
Abstract: In this talk, I am going to tell you how we at JetBrains use anonymous user logs to train machine learning models that are used in all IntelliJ-based IDEs to improve code completion. I am going to describe the pipeline that allows us to continuously perform A/B tests selecting the best models and keep our models up to date with the current trends in user behavior, while not violating the users’ privacy. The talk will conclude with the lessons learned and some takeaways for the research community.
Slides: docs.google.co...
Bio: Timofey Bryksin leads the “Machine learning methods in Software Engineering” research lab at JetBrains Research, aiming to apply data-driven methods in the software engineering field. Before that, he spent almost a decade working in industry as a software engineer and engineering manager. He received his B.Sc., M.Sc., and Ph.D. in Software Engineering from Saint Petersburg State University, Russia. His research interests revolve around software engineering tools and processes and how they can be enhanced with various types of data collected in this field.
Recorded on April 27, 2023.