Currently a freelance researcher with a keen interest on Machine Learning appraoches to Software Engineering. My interest lie in structured approaches to modeling source code, program synthesis, understanding defect prediction, and the testing oracle problem. I got my PhD from University College London where I worked under the supervision of Dr Earl T. Barr on improving project health by applying Machine Learning to problems from Software Engineering. Some of my projects are listed on this page.
My CV can be found here.
Aide-memoire: A tool to link issues and pull-requests in an online fashion by predicting which issues(PRs) relate to other PRs(issues). It makes use of a Mondrian Forest model that should be trained on a project before it can make predictions. It is composed of a backend(GitHub Link) and a Chrome plug-in to interface with the backend(GitHub Link)
POSIT: A tool that makes use of a CRF-biLSTM model to segment and tag text that mixes English and code snippets. It was trained on a combination of C code and StackOverflow. Project Page
Flexeme: A tool that untangles commits into atomic patches using graph kernel similarity and agglomerative clustering. It was validated on an artificial corpus of tangled commits for 9 C# projects. Project Page
Partachi, P.-P., Treude, C., Dash, S. K., & Barr, E. T., POSIT: Simultaneously Tagging Natural and Programming Languages. In 42nd International Conference on Software Engineering (ICSE ’20). Seoul, Republic of Korea: ACM., May 2020. Project Page
Partachi, P.-P., Dash, S. K., Allamanis, M., & Barr, E. T., Flexeme: Untangling Commits Using Lexical Flows. In 28th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering, (ESEC/FSE 2020). Sacramento, California, United States; ACM, November, 2020 Project Page