PhD Researcher · NIT Warangal

Vishal Kumar Swain

Formal Methods & AI for Software Verification

I combine formal methods with machine learning and LLMs — neuro-symbolic SMT solving over bit-vector logic, CBMC-based bounded model checking and MC/DC coverage analysis of safety-critical embedded C/C++, and LLM fine-tuning for coverage-driven test generation. I aim to apply this to formal verification and AI-assisted EDA.

IEEE ICST2026 tool paper
SMT-COMP2026 solver entry
14coverage criteria automated
9.5M.Tech CGPA

Research & Competitions

SMT-COMP 2026

NeuroSym: Neuro-Symbolic SMT Solver

Co-developed a solver that combines machine-learned guidance with symbolic SMT reasoning over bit-vectors, with Lucas Cordeiro's group at the University of Manchester. Entered in the International SMT Competition 2026.

SMT-LIBQF_BVNeuro-SymbolicPyTorch
IEEE ICST 2026

KSERESNET: Test Prioritization for Self-Driving Cars

An SE-attention 1D-ResNet that ranks simulation-based test cases from road-waypoint kinematics — APFD 0.62 vs. 0.50 for random ordering, with a 100% completion rate across 400 configurations.

Deep LearningResNetAutonomous Driving
Ongoing research

LLM-Driven Test Generation for Coverage Closure

Benchmarked 7 LLMs on 42 programs (coverage, mutation score, cost, latency, hallucination rate) and tested generalization on 65 GNU Coreutils programs. LoRA-fine-tuned StarCoder2-7B and fixed a severe post-fine-tuning drop with a chat-format-aligned training strategy. Built a coverage-guided feedback loop that finds uncovered branches after each round and re-targets them.

StarCoder2LoRA / PEFTHugging FaceMutation TestingvLLM

Experience

Jun 2026 – Present

Junior Research Fellow — Formal Software Verification

DRDO
  • Built an automated CBMC-based formal coverage pipeline covering 14 criteria, including MC/DC and mutation testing, with auto-instrumentation, verification-harness synthesis and a Python GUI.
  • Cut peak memory from 12–16 GB to ~4 GB by restructuring bounded model checking into per-file analysis with retry logic, removing out-of-memory failures on large codebases.
  • Led formal verification of safety-critical embedded C/C++ software, designing per-file stubbing strategies to verify tightly coupled modules in isolation.
Aug 2025 – May 2026

Junior Research Fellow

NIT Warangal
  • Contributed to TrustInn, a multi-language verification framework (bounded model checking, KLEEMA, KLEE-TX symbolic execution) for C, Java, Python, Rust and Solidity.
  • Built GAN- and LLM-based test-generation modules to maximize branch and line coverage, and containerized the tool with Docker.
Jul 2023 – Jan 2025

Lecturer & Teaching Assistant

GIET University, Gunupur
  • Taught core CS lectures and labs; mentored final-year ML and software capstone projects.

Skills

Formal Methods

CBMCBounded Model CheckingSMT (SMT-LIB, QF_BV)KLEETracer-XKLEEMAMC/DCMutation TestingAFL Fuzzing

AI & ML

PyTorchTensorFlowScikit-learnNeuro-Symbolic ReasoningGANsLearning to Rank

LLM Engineering

Hugging FacePEFT / LoRATRLvLLMOllamaLM Eval HarnessOpenAI / Anthropic APIs

Languages & Tools

PythonCC++RustSQLSolidityDockerGitLinux

Publications

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

Full list on Google Scholar.

Education

Aug 2025 – Present

PhD in Computer Science

Software Testing & Verification · NIT Warangal

2025 · CGPA 9.5 / 10

M.Tech in Computer Science

GIET University

Contact

Let's work on verification together.

Open to research collaborations in formal methods, AI for verification and AI-assisted EDA.