Hi, I'm Deeptendu Santra. Welcome to my thinking space.

I'm a (LLM) engineer passionate about the intersection of math, physics, and software. My primary focus is on building local, verifiable and trustworthy AI systems and I love quantifying "vibes" of a LLM.

I'm currently building applications to streamline my daily workflow and exploring new ways to interact with LLMs. [Read my manifesto here.]

I am driven by the belief that computational tools can significantly accelerate scientific discovery. During my time at the university, I worked on accelerating force and energy computations in organic compounds using GNNs. More recently, my research has shifted to gravitational waves, with a paper available on arXiv.

I am a strong advocate for Open Science and Software. I have been a previous Google Summer of Code (GSoC) contributor and mentor. I am always looking for opportunities to contribute to open-source projects, with a future focus on the PyTorch ecosystem and LLM compiler codebases.

I am currently seeking roles in LLM optimization and the development of verifiable AI systems. I love talking about LLMs, why OOP is bad, type theory, and tech-twitter drama. Feel free to reach out to me on X or you can try rickrolling me at hi@[thiswebsite].