Chennai, India
Software Development Engineer
I enjoy staying on top of what's new in tech and putting it to work — always learning something exciting, and always building something with it.
My curiosity has always lived one layer below the screen — in the routing, the queries, and the quiet machinery that decides whether an app feels instant or broken. A year at Cestrum took that curiosity into production: building APIs for a community of thousands taught me restraint, securing them taught me healthy caution, and the bugs that only appear under real traffic taught me more than any tutorial could.
My habit is to think in trade-offs before thinking in code. Competitive programming taught me to ask "what does this cost?" before writing a single line, and system design keeps asking the same question at a larger scale — what to cache, what to queue, and what to let fail gracefully.
My projects usually begin as complaints. Code reviews felt slow and inconsistent, so I built MergeMind, where five specialised AI agents split a pull request, review it in parallel, and return one coherent verdict. Incident response kept scattering across Slack threads and GitHub issues, so I'm building IncidentOS to bring triage into a single, reliable place.
Along the way I've earned my stripes as an AWS Solutions Architect and an Oracle Certified Professional. I treat them as a floor, not a ceiling — proof of the fundamentals, while the real learning happens in whatever I'm building next.
Vellore Institute of Technology, Chennai
2023 — 2027
CGPA : 9.1
College taught me consistency as much as computer science. Through coursework in systems, databases, and networks, I kept one simple routine: understand the concept in class, then build something with it before the week ended. That habit — steady, unglamorous, repeated — is where most of my growth has come from.
Relevant coursework
Backend Intern
Oct 2025 — Jun 2026
Chennai, India
Worked on the backend of a community platform serving 2K+ users.
A pull-request reviewer that works the way a good team does. A planner agent reads the diff and delegates to security, logic, and style specialists running in parallel, then a synthesiser folds everything into one clean, deduplicated review comment on the PR.
An incident-management platform that meets engineers where they already are — Slack and GitHub. AI-assisted triage sits on a hardened Express and PostgreSQL backend, with schema-validated APIs and rate limiting built in from day one.
Edit, save, and your Go server is already restarting. A self-contained live-reload CLI built directly on fsnotify — debouncing, glob filtering, and careful process lifecycle management, with no third-party hot-reload frameworks in sight.
Private contact discovery with context. Extends the FNP'04 Private Set Intersection protocol so identifiers only match when both the ID and its scope align — 'Alice at work' stays distinct from 'Alice in personal', and neither side learns anything else.
