Akarsh Simha


Brief Bio

I am in my late 30s and presently live in the San Francisco Bay Area in California and am building AI agents for a company focused on sustainable energy and home-improvement called GoodLeap. I am educated as a physicist and have chosen a career in software, machine learning and artificial intelligence.

Even though I did not formally study computer science/engineering save for a few classes, it has been a passion of mine since middle school. Association with peers studying computer science during undergraduate college, the Google Summer of Code program, as well as working on the open source project KStars for many years honed my skill with computers before I started my career as a software and ML engineer.

Before moving to the silicon valley, I got a PhD in Physics from the University of Texas at Austin. I was advised by Phil Morrison and co-advised by Mark Raizen. My thesis was on theoretical and experimental aspects of Brownian Motion in Liquids. I was very fortunate to work both on theoretical and experimental aspects of the same problem, picking up a diverse set of skills in the process. One of our results was published in the Science magazine.

I then worked at Apple, Inc. in several different aspects of machine learning – a training framework (thought of as a compiler), training/evaluating/deploying several neural-network models on the iPhone and Mac, building model-in-the-loop data annotation tools, contributing to model-training infrastructure, defining and supervising the delivery of complex training data for novel models, and on a model-deployment framework. In addition, I had an opportunity to rotate for 3 months into an LLVM-based GPU compiler team contributing to performance optimizations. One of the features that I am proud of contributing to was the static subject lifting feature on iOS released in 2022 that can be used to make stickers, that was highlighted in Apple's WWDC keynote as well as in an Apple machine learning blog post.

I indulge in astronomy as a hobby, tinker around with my Linux machines, learn math and physics, go hiking / car camping / road tripping / photographing, build small DIY things, develop software as a hobby.

The rest of this website is for sharing my hobby work rather than my professional work. If you're further interested in my professional contributions, see my Resume or the other links in the header.

Publications and Patents

My professional publications are also listed on Google Scholar.

Patents

Experiments on Brownian Motion

Low–Reynolds number fluid dynamics

Other physics

Amateur Astronomy

You may contact me via e-mail, the username is akarsh and the domain is @kde.org