Raghuram Shankar

I build machine learning models for batteries, and just as importantly, models that know how much they don’t know.

At Elysia I work at the intersection of probabilistic machine learning, Bayesian state estimation and battery prognostics. My background sits where deep battery physics meets modern probabilistic ML: PyTorch, JAX and Pyro/NumPyro on one side, electrical, thermal and ageing models on the other. Before Elysia I built battery state estimation software for EVs at NEVS, and I studied electrical engineering at Chalmers.

Outside work I follow the markets closely. Risk management and portfolio construction are a longstanding hobby, and I automate whatever I can with Python. Lately I’ve been teaching myself new topics with LLMs and publishing them as interactive explainers below.

Blog