I’m a PhD student at the National University of Singapore, where I work in the Programming Languages and Software Engineering (PLSE) Lab within the Trustworthy Software Systems group, advised by Professor Abhik Roychoudhury. My research spans automated program repair, automated verified evolution of software systems, and AI-assisted software security.

Research Directions

Automated Program Repair. Building systems that automatically find and fix bugs in real-world code, with infrastructure for evaluating them fairly.

• AutoCodeRover: Agentic Program Repair for SonarQube Issues (FSE 2026) — M. Mirchev, R. Shariffdeen, H. Ruan, Y. Zhang, A. Roychoudhury
• Assured Automatic Programming via Large Language Models (arXiv 2024) — M. Mirchev, A. Costea, A. K. Singh, A. Roychoudhury

LLM-Driven Fuzzing & Security. Using LLMs to understand code and protocol semantics well enough to find and fix real vulnerabilities, while hardening the software supply chains behind AI systems.

• Fixing Security Vulnerabilities with Agentic AI in OSS-Fuzz (ICSE 2026) — Y. Zhang, J. Wang, D. Berzin, M. Mirchev, A. Roychoudhury
• Large Language Model guided Protocol Fuzzing (ChatAFL, NDSS 2024) — R. Meng, M. Mirchev, M. Böhme, A. Roychoudhury

Alongside my PhD, I co-founded AutoCodeRover, an NUS spin-off building an autonomous AI agent platform for automated software development. I served as its CTO from 2024 until its acquisition by Sonar in February 2025. I now continue this work at Sonar as a Research Consultant, after first working there as a Research Engineer, focusing on agentic program repair and AI-assisted issue resolution.

Before NUS, I completed my B.Sc. in Computer Science and Engineering at Eindhoven University of Technology (cum laude) and worked on compiler and grammar tooling (ANTLR) as well as full-stack engineering.

You can find a full list of my papers on the publications page or on my Google Scholar profile.

Fun Fact

I’ve tried my hand at less conventional sports, including kendo and archery, but swimming remains my go-to.