SAC 2026
An Empirical Study on Spectrum Based Fault Localization Heuristics
The 41st ACM/SIGAPP Symposium on Applied Computing, Poster Paper Track
PhD student, Department of Computer Science and Engineering, IIT Kanpur
I work on finding the cause of bugs in software: fault localization, program analysis and automated testing, and where language models fit into that work. My advisor is Prof. Subhajit Roy.
I study how software can be made easier to understand, debug and test, using empirical studies, program analysis and AI-assisted tools.
Peer-reviewed work in software engineering. The BibTeX button copies the citation to your clipboard.
SAC 2026
The 41st ACM/SIGAPP Symposium on Applied Computing, Poster Paper Track
ASE 2025
40th IEEE/ACM International Conference on Automated Software Engineering, Tool Demonstration Track
ASE 2023
38th IEEE/ACM International Conference on Automated Software Engineering, Research Track
Research tools and open-source software I’ve built.
Personal project
An open-source personal AI system and agent framework that runs on your own machine. It senses your system and workspace, remembers conversations, and lets agents use tools only after a permission check. Generating and verifying new tools from a description is planned but not built yet.
SAC 2026
A stochastic sampler for fault spectrums. Spectrum-based fault localization is effective and scalable, but its metrics are usually compared under limiting assumptions. ODIN samples spectrums from carefully chosen control parameters, so the comparisons hold up statistically. With it I studied 37 popular metrics, generalized several earlier findings, and found new equivalence relations that hold across experimental configurations.
Code not released yet
ASE 2025
A tool for diagnosing faults in Android apps. It pairs an evolutionary test generator with a statistical fault localization engine. The generator supports configurable fitness functions, such as coverage and diagnosability metrics, and the localization module supports metrics including Ochiai, Tarantula and Barinel. On 20 open-source F-Droid apps it reduced debugging effort substantially and ranked more faults among the top candidates than random testing.
ASE 2023
A program analysis framework for teaching graduate courses. It keeps the code base small, uses one intermediate representation, and has a modular design for plugging in new algorithms. It covers dataflow analysis, abstract interpretation, symbolic execution, fuzzing, evolutionary test generation and spectrum-based bug localization, and has been used in several course offerings.
I’ve been a teaching assistant for these courses at IIT Kanpur.
I’m happy to talk about research collaborations, debugging tools, software engineering, or AI-assisted development.