Who am I?

I like building random stuff but rarely write it down in a way anyone else can follow. Mostly because I’d rather spend time doing things than making perfect docs. Once you start documenting, it’s gotta be spotless.

But hey, the time came when I had to do it, since it’s easier to show off the things I’m proud of — so here we are.

If you stumbled here by chance, I’m Kesava Prasad Arul. I do all sorts of random things, mostly around GPUs, profiling, and making stuff run faster.

Right now, I’m working on my Master’s thesis at TUM and Bosch. I’ve had quite the journey at Bosch too — starting back in Jan 2018, then Aug 2018, Jan 2021, Aug 2024, Oct 2024, Jul 2025, and Nov 2025. No, that’s not a glitch or a repeat — just some serious loyalty.

I was a Powertrain Engineer until 2023, working with diesel engines and thermodynamics. That meant testing a bunch of fine cars — Audi, BMW, Mercedes/Daimler, FCA/Stellantis, Ferrari, Porsche, and more — making sure their emissions were on point. It was a blast, 10/10 would recommend.

In 2023, I switched hard into hardware optimization for ML on ECUs. These things run on significantly tight resources — your phone is probably 100 times faster. That challenge hooked me. It made me a TinyML engineer and pushed me to learn the hardware side of ML.

After that, I jumped back into academia at TUM, for Electronics and Communication Engineering, where I got to understand how hardware really works — FinFETs and it's derivatives, why transistor scaling below 2nm is so tough, and all that jazz. Suddenly, processors made sense. What looks like magic from a software-only view is actually super logical once you know the hardware side. It's a whole conversation to talk about, really.

During my studies, I kept coming back to Bosch, hopping between different business units — Heating Units (Home Comfort), Brakes (Vehicle Motion), LLM/RAG pipelines (Vehicle Motion), and now I’m slowly moving into AI/ML research groups focusing on function calling, linear optimization, and multi-constraint optimization for CAD designs, in collaboration with the greatest Corporate Research team I've seen. It’s been a wild ride.

I’m aiming to wrap up my thesis by May 2026. Hopefully, I’ll find something just as exciting to dive into next.

On the side, I’m part of this awesome student group called TUM.ai. I lead a cute little task force called Applied Accelerated Computing. Basically, I teach people why picking the right hardware for the right problem matters. GPUs aren’t always the answer, and neither are the best CPUs. You have to understand the problem and figure out where hardware can really help. That’s what I’m passionate about sharing.

Besides that, I’ve done a bit of everything in the community — marketing and photography, software development (helped build TUM.ai), UX/UI (yeah, I made the homepage and animations with my rusty CSS/HTML skills), recruitment, and even helping startups grow into million-euro businesses. I also helped another person from my team to start his own task-force which I'm incredibly proud of.

Beyond this, I had phases of hackathon winning sprees. I stand at 26 podium spots in a total of 69 nice hackathons. I can get competitive sometimes, but it's mostly because the problems to solve are so damn interesting.

I’m always chasing the next problem that forces hardware and intelligence to meet in unexpected ways. If you're there too, hit me up and we'll talk :) If not, but you wanna hang here, welcome to my journey to the convergence.

Contact: me@kesava.lol