About
I grew up in Karnataka, India, and studied computer science at NIT Karnataka. Somewhere in those years I picked up the conviction that still drives me: the whole stack is knowable, from the transistor to the product, and the interesting problems live at its limits.
By my final year I was doing research: a capsule-network architecture for segmenting retinal fluid that cut trainable parameters by half compared to U-Net. It was published in Springer's Medical & Biological Engineering & Computing, and it taught me that a tight constraint, fewer parameters on less compute, is a better teacher than a big budget.
At Samsung R&D I spent a year and a half about as close to hardware as software gets. The Night Mode multiframe pipeline I helped architect runs on millions of Galaxy devices: C++, Neon SIMD, GPU shared memory, quantized models on NPUs. Down there, latency is measured in fractions of a millisecond and paid for in physics; there is no patch Tuesday for a camera. I learned what optimization really means when the machine is small and the constraint is real.
Having seen how far software can be squeezed at the small end, I wanted the opposite limit: what happens when it has to span the globe. After a Master's at USC and a summer at Skyworks making TensorFlow Lite for Microcontrollers faster (the fixes are upstream), I joined Apple in Austin. I build the retail platform behind every Apple Store on the planet, the systems real customers touch when they book an appointment or walk in to buy something. At that scale the hard problems change shape: peak traffic, availability, reliability, the non-functional limits you only meet when the whole world shows up at once. Event-driven systems on Kafka and Temporal, holding a 99.99% uptime bar.
That's the arc this site records: the same engineer chasing the limits of software at both ends. Silicon, where a millisecond is an eternity. Planet scale, where a minute of downtime is one. On my own time I place bets on what comes next, agents and local AI mostly. I'm building toward founding something; the bets are how I find out what. This site is where I keep the record honest.