Bengaluru, Karnataka, India
I'm a software engineer currently at Google, working on Android Studio - mostly on the performance tooling side, and more recently on AI-powered features for Android developers. Before this I was at Microsoft for three years on the Extensions store, where I worked on a lot of different things: the store frontend, backend services, search, database performance, security and Edge client. It was a good mix of breadth and depth. Before that, Qualcomm - embedded systems, mostly C++ work on a broadcast platform, which gave me a decent foundation in thinking about performance and low-level constraints. Right now I'm most interested in the space where systems knowledge meets applied AI - specifically, how you can use LLMs to make complex technical domains like performance analysis less opaque for developers. Happy to connect if any of this overlaps with what you're working on.
Extensions Store redesign - 35% improvement in badged extension conversion, 20% faster page loads, 25% reduction in bounce rate. Gold Accessibility Award recipient. Multi-region and multi-locale search infrastructure across 240+ markets and 84 locales, improving conversion from 30% to 47%. x-Team collaboration to further enable consistent search experience across Extensions Store and Extensions Store on Sidebar. Extension Validation Service rewrite into an orchestration pattern for improved performance and stability. Peak DB write throughput reduced from ~85% to ~25% through a batching redesign. Security across Extension Store services - WAF configuration, rate limiting, and passwordless authentication. Browser-native C++ APIs for targeted feature rollouts. Primary point of contact for regulatory data and telemetry requirements. Team-level live-site review processes, MTTx dashboards, and monitoring infrastructure. ⭐ Star Award for Execution.
eMBMS platform - 42% reduction in binary size and 20% reduction in CPU usage through compiler optimizations, code fixes, and boot-up improvements. Modularization of server-client middleware with cleaner abstractions and a facade layer. RTP stream playback support for broadcast data. Crash-dump persistence and remote debugging on low-memory devices for improved debuggability.
Winner, Qualcomm IdeaQuest. Built a classification algorithm to automate defect routing and reduce manual triage, shipped with a GUI for log analysis.