Kohei Dozono, M.Sc.
Technical University of Munich
Informatics 4 - Chair of Software & Systems Engineering (Prof. Pretschner)
Postal address
Boltzmannstr. 3
85748 Garching b. München
- Phone: -
- kohei.dozono@tum.de
About Me
I joined the Chair of Software and Systems Engineering (Prof. Dr. Pretschner) as a Ph.D student in 2024. My research focuses on security testing, particularly static application security testing and fuzz testing. I am also interested in applying large language models (LLMs) to enhance security testing methodologies.
Prior to joining the program, I completed my Master's degree in Informatics at TUM. My bachelor's journey began at the Kitami Institute of Technology in Japan. I transferred to pursue a dual degree program, earning a Bachelor of Computing from UWE Bristol and a Bachelor of Computer Science from Taylor's University.
Thesis Supervision
If you are interested in working with us or need supervision for your bachelor's or master's thesis, please feel free to contact me! If you have an idea related to my research interests, you're also welcome to present it to me for discussion on possible supervision.
For the open position below, please send me an email with your CV and transcripts attached, and include a brief motivation (max 5 sentences) directly in the email.
--- Open ---
--- Ongoing ---
| Seed Optimization for Directed Greybox Fuzzing in Commit-Change Testing | Master's/Bachelor's Thesis |
--- Completed ---
| (2025) Fuzz Smarter, Not Harder: Simplifying Fuzzer Use with LLMs (with Infineon Technologies AG) | Project Work |
| (2025) Identifying the Cone of Influence for Commit Fuzzing. | Master's Thesis |
| (2024) Large Language Model Based Security Analysis of Source Code Written in Proprietary Languages (with CQSE GmbH) | Master's Thesis |
Teaching
| Semester | Course |
| Winter Semester 24/25 | Seminar - Softwarequalität |
| Summer Semester 25 | Lecture - Security Engineering |
Publications
- Dozono, Kohei; Gasiba, Tiago Espinha; Stocco, Andrea: Large Language Models for Secure Code Assessment: A Multi-Language Empirical Study. 7th IEEE/ACM International Workshop on Deep Learning for Testing and Testing for Deep Learning (DeepTest), DeepTest 2026.
- Dozono, Kohei; Engesser, Jonas; Hummel, Benjamin; Roehm, Tobias; Pretschner, Alexander: Should We Evaluate LLM Based Security Analysis Approaches on Open Source Systems?, 40th IEEE/ACM International Conference on Automated Software Engineering, ASE 2025.