01 / HelloA little about me ↗
02 / NeurIPS, 2025First place, with my brother ↙
03 / UC San Diego, 2023The people along the way ↙I work on language models at eBay. Before that: UC San Diego, Texas Instruments, and NIT Karnataka.
I also write, make things, and sometimes turn an idea into a film.
From an electronics lab in Karnataka to chip design, graduate school, and AI research. The people, projects and photographs along the way.
My brother Chin and I won first place in the NeurIPS 2025 Embodied Agent Interface Challenge, as team AxisTilted2. We built language-model agents that reason about physical tasks in simulated environments.
Joined eBay’s Knowledge Extraction team in April 2024. Researching and building language-model systems for information extraction. Promoted in October 2025.
Completed an MS in computer science at UC San Diego. Worked with Julian McAuley’s group on AI music and language models, and TA’d recommender systems and data mining.
ZINify received an Honorable Mention at the UIST Student Innovation Contest. Experiments with language models and visual storytelling.
Accepted into UCSD’s StartR Rady accelerator with Glyp, a writing assistant for novelists. The project did not make it to market; I wrote about what went wrong.
Summer research with eBay’s Knowledge Extraction team, working on information extraction at scale.
Won the eBay University Machine Learning Challenge by extracting named entities from product titles using DeBERTa V3 and K-fold ensembling. The result, announced in January 2023, led to the summer research internship.
Started the MS at UC San Diego after three years in chip design.
ASIC digital design at Texas Instruments, 2019–2022: physical design, RTL, and getting actual chips out of the door.
Graduated in electrical and electronics engineering. I got into deep learning through Kaggle, earning silver and bronze medals. My thesis on power-quality classification won a best-paper award at IEEE DISCOVER; the Amateur Astronomy Club gave me a place to explore space and astronomy.
My brother and I used a language model and the competition’s own evaluator to generate training data and win the NeurIPS 2025 Embodied Agent Interface Challenge.
An essay about how advances in AI and cheaper robot hardware could bring capable robots into everyday life. Also adapted into an animated film.
The story of building a tool for novelists through UCSD’s StartR accelerator, and why it never made it to market.
Walk or fly inside a rotating space habitat, with cities, lakes and a landscape that curves over your head. An interactive O’Neill cylinder.
Explore the space habitatWhat would Claude do with a quiet life as a bee? Watch it make decisions in a garden simulation.
Open the bee simulationAn interactive simulation of dismantling Mercury to build solar collectors in orbit around the Sun.
Run the solar-collector simulation
Watch the film ↗How to Please a Capricious God — a painted allegory about agents, their tests, and the eye watching over them.
Enter the habitat ↗Another Sky — walk and fly inside an O’Neill cylinder. Runs in your browser.
My essays, turned into painted films about AI and what comes next.
