Compilers make code fast. I study when fast means unsafe.
I’m a PhD student in Computer Science at Baylor University, studying how compiler optimizations interact with undefined behavior, security bugs, and programmer assumptions in C programs.
int read_len(struct pkt *p) { int len = p->len; if (p == NULL) // safety check return -EINVAL; return len;}p was already dereferenced, the optimizer assumes it can't be null and deletes the check. The same pattern made a Linux kernel bug exploitable in 2009 (CVE-2009-1897).
About
I work with Dr. Mengfei Ren at Baylor University on compiler correctness and security. Compilers sit underneath nearly every piece of software, so a single wrong optimization can break or weaken thousands of programs that were written correctly. My research asks which optimizations go wrong, why, and how we can catch them before they ship.
Before graduate school I spent five years as a security engineer, first in R&D at Quantei and then at Wazuh, where I wrote technical content for its open-source security platform. Those years taught me to focus on problems that show up in production, not just in benchmarks.
I've studied in Sri Lanka, Russia and the United States, and I speak Bengali and English.
Research
Bugs introduced by compiler optimization
When does making a program faster make it incorrect?
I study real optimization bugs in GCC and LLVM to understand which passes fail and how, then build techniques to detect them automatically in C compilers.
Security consequences of compilation
Can a correct compiler still make code less safe?
Optimizations that are legal under the C standard can remove safety checks or expose memory errors. I work on static analysis and security-preserving optimization so performance doesn't come at the cost of safety.
AI-assisted security analysis
How can machine learning help teams ship secure code?
I've built ML-based security analysis for DevSecOps pipelines and earlier worked on deep-learning detection of deepfake audio.
I'm also interested in the security of large language models and multi-agent AI systems.
Background
Education
- PhD, Computer Science (Cybersecurity)Baylor University, Waco, TX, 2025–present
- PhD studies, Computer EngineeringUniversity of Alabama in Huntsville, 2024–2025, transferred to Baylor
- MS, Security and Network EngineeringInnopolis University, Russia, 2022–2024
- BS, Information Technology (Cybersecurity)SLIIT, Sri Lanka, 2017–2021
Experience
- Graduate Research AssistantBaylor University, 2025–present
- Graduate Research AssistantUniversity of Alabama in Huntsville, 2024–2025
- Cybersecurity Engineer, Technical ContentWazuh Inc., 2022–2024
- Cybersecurity Engineer, R&DQuantei Australia, 2019–2022
- AWS Cloud AmbassadorAmazon Web Services, 2019–2020
Honors
- Highest PerformerWazuh Inc., 2023–2024
- Innopolis University Scholarship2022–2024
- SAARC Scholarship2017–2021
Skills
- Languages
- C, Python, Bash, PowerShell, Swift, x86 assembly
- Machine learning
- Deep learning, NLP, reinforcement learning
- Infrastructure
- Git, Docker, Kubernetes, Ansible, Terraform, AWS, GCP, Azure, LaTeX
Contact
I'm glad to talk about research collaborations, opportunities in security, or anything where compilers meet security.
Abdullah_Noman1@baylor.edu