Welcome #

I ask whether we can trust AI that writes code, and try to give the field the tools to answer. I am a postdoctoral researcher at Delft University of Technology, following a PhD in the Software Engineering Research Group (defence December 2026). My research covers the security and evaluation of large language models and AI agents for software engineering: what code models memorise and leak, how they behave when deployed, and how to measure what agents can reliably do. Previously, I obtained my Masters in Computer Science and Engineering - with a specialisation in Cybersecurity.

Email: a.al-kaswan[at]tudelft.nl

Recent Publications #

Here are some of my recent research contributions:

• Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture The Flag Challenges (AIWare'26) Partial-credit benchmarking of LLM agents on realistic CTF challenges

• Model See, Model Do? Exposure-Aware Evaluation of Bug-vs-Fix Preference in Code LLMs (MSR'26) Investigating how exposure influences the propensity of LLMs towards bugs and fixes

• Code Red! On the Harmfulness of Applying Off-the-shelf Large Language Models to Programming Tasks (FSE'25) A comprehensive framework for assessing the potential harmfulness of LLMs in software engineering

• Traces of Memorisation in Large Language Models for Code (ICSE'24)
  Exploring memorisation in code language models and data extraction vulnerabilities

• The (ab)use of Open Source Code to Train Large Language Models (NLBSE'23)
  Discussing security, privacy, and licensing implications of LLM training

View All Publications →

Other Contributions #

Editorial & Review Activities #

• ACM Transactions on Software Engineering and Methodology (TOSEM) Reviewer (2026)
• Science of Computer Programming (SCP) Reviewer (2026)
• IEEE Transactions on Software Engineering (TSE) Reviewer (2026)
• 16th Intl. Conference on the Quality of Information and Communications Technology (QUATIC'26) Program Committee - Security and Privacy Track (2026)
• ACM Transactions on Software Engineering and Methodology (TOSEM) Reviewer (2025)
• 18th Intl. Conference on the Quality of Information and Communications Technology (QUATIC'25) Program Committee - Security and Privacy Track
• ACM Transactions on Software Engineering and Methodology (TOSEM) Reviewer (2024)
• Science of Computer Programming (SCP) Special Issue Executive Guest Editor
  Executive editor of the special issue for NLBSE'24 in the Science of Computer Programming Journal 📖 NL-based Software to Support SE Processes
• 3rd Intl. Workshop on NL-based Software Engineering (NLBSE'24) Program Committee

Conference Organization & Service #

• The Poisoned Chalice of LLM Evaluation Competition Co-Chair (FSE'26) 📄 Competition Report
• 4th Intl. Workshop on NL-based Software Engineering (NLBSE'25) Tool Competition Co-Chair 📄 Competition Report
• 3rd Intl. Workshop on NL-based Software Engineering (NLBSE'24) Tool Competition Co-Chair 📄 Competition Report

Conference Participation #

• ICSE'23 Doctoral Symposium
• Dutch National Symposium on Software Engineering (SEN) Talk (2023) Lightning Talk Titled: Co-Pilot the Open Source Laundromat
• ICSE'23 Student Volunteer

Awards 🥇 #

• ACM SIGSOFT Distinguished Paper Award
  For the paper: How Much Do Code Language Models Remember? An Investigation on Data Extraction Attacks before and after Fine-tuning at MSR'25 📄 Paper

• Training Data Extraction Challenge - First Place
  SATML'23 Training Data Extraction Challenge
  📄 Competition Report

• Code Comment Classification - First Place
  NLBSE'23 Code Comment Classification Competition
  📄 Competition Report