{"schema_version":"1.0","variant_id":"public-profile-v1","request_id":"ca201207ad92aff096d67085cc0448da","content_as_of":"2026-05-18","source":{"name":"Jonathan Katzy","url":"https://jkatzy.nl/","language":"en"},"profile":{"name":"Jonathan Katzy","role":"PhD candidate in Trustworthy Code Models for Software Engineering","affiliation":"Software Engineering Research Group, Delft University of Technology","biography":"My research focuses on making code language models more trustworthy for software engineering. I study how these models can be explained, evaluated, maintained, and designed inclusively, using empirical methods and human-in-the-loop approaches that keep developer needs, real-world software engineering practices, and human values at the center.","interests":["machine learning","natural language processing","software engineering","programming languages"],"links":{"github":"https://github.com/jkatzy","google_scholar":"https://scholar.google.com/citations?user=mmy9FeIAAAAJ\u0026hl=en","linkedin":"https://www.linkedin.com/in/jonathan-katzy-568263138/","research_group":"https://se.ewi.tudelft.nl/"}},"results":[{"type":"publication","title":"An Exploratory Investigation into Code License Infringements in Large Language Model Training Datasets","url":"https://jkatzy.nl/publication/code-license-infringement/","date":"2024-08","authors":["Jonathan Katzy","Razvan-Mihai Popescu","Arie van Deursen","Maliheh Izadi"],"summary":"An empirical study of code-license inconsistencies in training datasets for large language models. Across 514 million code files, the study found substantial overlap with strong-copyleft code and recommends better dataset creation and management practices."},{"type":"publication","title":"Language Models for Code Completion: A Practical Evaluation","url":"https://jkatzy.nl/publication/code4me/","date":"2024-07","authors":["Maliheh Izadi","Jonathan Katzy","Tim van Dam","Marc Otten","Razvan-Mihai Popescu","Arie van Deursen"],"summary":"A quantitative and qualitative evaluation of public code-completion models using more than 600,000 real completion interactions from over 1,200 Code4Me users across twelve programming languages."},{"type":"publication","title":"On the Impact of Language Selection for Training and Evaluating Programming Language Models","url":"https://jkatzy.nl/publication/impact-of-language-selection/","date":"2023-08","authors":["Jonathan Katzy","Maliheh Izadi","Arie van Deursen"],"summary":"A CodeBERT-based study of programming-language representation similarity and its implications for selecting diverse languages when training and evaluating code models."}],"notice":"agent-name and purpose are self-reported and are not treated as verified"}
