I combine formal methods with machine learning and LLMs to make software verification smarter and more scalable.
- 🧠 Neuro-symbolic SMT solving over bit-vector logic — entrant at SMT-COMP 2026
- 🛡️ CBMC-based bounded model checking and MC/DC coverage for safety-critical embedded C/C++ — JRF at DRDO
- 🤖 LLM fine-tuning (LoRA, StarCoder2) for coverage-driven test generation
- 🚗 ML-based test prioritization for self-driving cars — IEEE ICST 2026 tool paper
- 🎯 Next: formal verification and AI-assisted EDA
| Project | What it does | Venue |
|---|---|---|
| NeuroSym | Neural-symbolic SMT solver (SMT-LIB, QF_BV), with Lucas Cordeiro's group, Univ. of Manchester | SMT-COMP 2026 |
| KSERESNET v2 | SE-attention 1D-ResNet for SDC test prioritization — APFD 0.62 vs 0.50 random | IEEE ICST 2026 |
| ARSF | Adaptive Risk-Semantic Fusion: KSERESNET + Qwen3-8B for SDC test prioritization | Research |
| VeriRust | Automatic Kani harness synthesis for formal verification of Solana Anchor smart contracts | Research |
| CAST | Code Attribution Stability Testing — artifact | ISEC 2027 (submitted) |
| NeuroSym-benchmarks | Buggy/correct C benchmark corpus from ESBMC's regression suite | Dataset |
- KSERESNET at the ICST 2026 Tool Competition: Self-Driving Car Testing Track — Tool paper, 19th IEEE International Conference on Software Testing, Verification and Validation (ICST 2026), Daejeon, South Korea. IEEE Xplore.
Open to research collaborations in formal verification, neuro-symbolic AI and AI-assisted EDA.