Hey, I’m Aditya.

I’m currently researching independently. I don’t confine my research or work to just one domain. I pick up problems that I find interesting and that could have a considerable impact. My work currently spans robotics, machine learning, and embedded systems. Recently, my ML research has focused on the coding abilities of language models. I’ve been working on coding benchmarks, checking whether they measure what they claim to, and developing coding harnesses and other evaluation tools. I’m also learning more about specification gaming and mechanistic interpretability, especially whether interpretability techniques can help design better benchmarks. I’m working on RL in EVs and trying to understand Formula 1.

Feel free to contact me regarding my work and collaborations.

A reminder I keep +