I’m a final year Computer Science student at VIT Chennai, and most of what I’ve learned so far has come from building things, breaking them, and figuring out how to fix them.
I spent about a year as a backend intern at Cestrum, where I worked on APIs for a community platform used by 5K+ people. I worked on posts, feeds, notifications, authentication, and role based access. Working with JWT and Google OAuth 2.0 taught me to be more careful about security. Replacing polling with WebSockets showed me how much of a difference real time features can make. I also worked quite a bit with MongoDB queries and indexes, which taught me that many performance problems are not as complicated as they first seem.
Before that, I was a DevOps intern at Comviva in Gurugram. I worked with AWS, Terraform, GitHub Actions, Docker, and Linux, mainly around infrastructure and deployments. That experience helped me understand the other side of backend development and made me think more about how the code I write actually gets deployed and runs in production.
More recently, I’ve been exploring AI agents and building projects around them. That’s where IncidentOS and MergeMind came from. Both started with me trying to improve workflows that felt repetitive and took more time than they should.
I’m also an AWS Solutions Architect and an Oracle Certified Professional, but I see certifications as a starting point rather than the main goal. Most of what I learn still comes from building something, running into problems, and figuring out how to solve them.
An incident response platform that never acts without a human saying yes. Seven agents triage the alert, correlate state across Slack, GitHub, Prometheus and Grafana, and draft a remediation plan — which then waits at an approval gate until someone signs off.
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.
"Build systems that outlast the sprint."