About — Ashwin Rohit Alagiri Rajan

I am Ashwin Rohit Alagiri Rajan, and I study Computer Engineering at the University of California, San Diego, in the Department of Computer Science and Engineering. I am advised by Prof. Hadi Esmaeilzadeh, and I work on computer architecture — specifically on acceleration for workloads that are outgrowing the machines they run on.

Research

Two threads run through my current work. The first is what an accelerator should look like when the workload will not hold still: Accelerator Polymorphism (ISCA 2026) asks whether a domain-specific architecture can be made to shift shape across the very different kernels a robot has to run, rather than being fixed at tape-out for one of them. The second is what an accelerator owes the people whose models and data pass through it: IroKnight (ISCA 2026) is an ownership-preserving neural accelerator for inference serving, aimed at the case where a model owner and a serving provider do not fully trust each other.

Earlier, I worked on the environmental cost of computing hardware rather than its speed. Reducing the Carbon Footprint of EdTech with Repurposed Devices (IGSC 2024) looked at whether retired consumer devices can carry real educational workloads instead of being replaced, and what that substitution actually saves. The publication list, with papers and BibTeX, is on the home page.

Teaching and writing

I helped build and release an open-source parallel programming curriculum, which I presented at the Qualcomm Platforms Symposium in 2024. I also write occasionally about the parts of systems programming that are underdocumented — a walkthrough of getting started with Metal C++, for instance. My graphics work includes a relativistic black hole ray tracer that traces null geodesics through curved spacetime, which was selected as a best final project in UC San Diego's CSE 168.

Elsewhere

The quickest way to reach me is by email; details and postal address are on the contact page.