Experience
Late 25 -
Building AI systems in stealth across data integration, retrieval/context, evals, observability, and backend/product infrastructure.
Nov 25 - Mar 26
At Tesla, I worked on Linux Software Platforms, the embedded Linux layer that vehicle AI and infotainment depend on. My work was mostly behind visible features: pieces of Grok's hands-free voice path, display performance, and protected phone/media projection across low-level services, graphics/audio pipelines, drivers, and hardware interfaces.
Jun 24 - Nov 25
At Simbe, I worked on Tally, a robot that has to navigate live retail stores and keep collecting useful inventory data. On a small robotics team, I moved across the C++/Linux foundations and fleet layer: autonomy and recovery, localization, sensor pipelines, edge performance, real-time telemetry/video, robot-cloud protocol work, and field debugging.
Sep 22 - Dec 23
In the Autonomous Robotic Manipulation Lab, I worked on robot learning for planning and control, advised by Tom Power and Prof. Dmitry Berenson. My piece was sample-efficient inference for MPC: task-space kernels for diverse Stein variational control in contact-rich manipulation.
Jun 22 - Aug 22
I interned with AWS as a research software engineer, building classical and RNN-based forecasting models for backbone capacity planning. I also worked on feature engineering and data infrastructure for network topology and telemetry signals.
Jan 22 - Apr 22
I was a software engineer fellow at Academic Innovation, working on GradeCraft. I built product features, worked through a Ruby migration, and improved backend paths around grade calculation and release workflows.
Jun 21 - Aug 21
I interned with Quadeye as a systems software engineer. I worked on low-latency trading systems, signal-processing optimization, and Linux performance tracing.
Michigan
At Michigan, I studied computers and mathematics seriously, with enough history, physics, writing, and film around the edges to keep the world bigger than the problem set.
EECS clicked when separate courses became one computing stack: bytes moving through memory, disks, networks, and devices; work being scheduled; state staying consistent; failures being contained. I kept the habit of measuring the path, naming the invariant, and knowing which layer owns the fix.
Honors math was hard thinking in public: small groups at boards, building proofs instead of receiving them. Analysis carried calculus onto manifolds, forms, Stokes, and de Rham; algebra, probability, combinatorics, and numerical linear algebra gave other languages for structure, symmetry, uncertainty, and stability. The best parts felt like research: model, conjecture, write clearly.
I taught along the way too: a SWE capstone, linear algebra for intelligent systems, Math Corps in Detroit, and grading across theory of computing, advanced + randomized algos, and cryptography.
Outside class, I spent time with Solar Car strategy, V1, HKN, and friends building or reading things.
Past
I grew up in New Delhi, India and graduated from Delhi Public School, R K Puram.
I had the best time of my life learning and building in tech with the people most passionate about it, as part of Exun Clan. I also served on the Student Council, was part of the quizzing and math clubs, and got a Red Tie Award for excellence in CS.
