Main research area involves applying machine learning to code processing; involved in several projects related to program repair, including but not limited to developing LLM fine-tuning methods for program repair, establishing executable code benchmarks, etc.
Education
PhD Student at KTH Royal Institute of Technology
Background
Research interests include machine learning on code. His PhD topic is differentiable program repair: translating programs into neural networks to perform repair at the numerical level. Additionally, he has developed supervised and self-supervised LLM fine-tuning approaches for program repair, created executable code benchmarks, evaluated the latest frontier models, and developed a tool to collect executable code datasets.