Shobhan

Have most of my past experience in large legacy physics solvers, plus generous exposure to Python, HPC/MPI and numerical methods, and some scientific ML.

Picking up the modern tooling for scientific computing, viz. OOP, JAX, and C++.
But otherwise, mostly learning more and more of the engineering craft behind agentic AI and develop some of my own.

Right now, exploring agentic AI tooling: harness and loop design, token economy, and efficient agent memory and retrieval.

Keen to trade notes on novel ways of getting real work out of agentic AI and on scientific computing, and the latest in SciML.