Agentic AI
HEP-Genesis Agent
A desktop and CLI research agent that drives science tools and runs compute locally or on ALCF and NERSC.
Computational Scientist · Argonne National Laboratory
Nesar Ramachandra
I build AI methods for scientific discovery: foundation models and agentic workflows for science, emulators with calibrated uncertainty quantification for expensive simulations, and synthetic sky catalogs from exascale cosmological simulations. My work is aimed at cosmology and astrophysics, from Rubin LSST and SPHEREx to the physics of dark matter.
Recent highlights
All highlights →
Datasets & simulations
SuperMock 3D
Fly through ~800K mock galaxies painted onto the HACC Last Journey simulation for SPHEREx.
Tutorials
Building Scientific Agents
Build a LangGraph multi-agent client that calls science MCP servers to reproduce published figures.
Emulators
Subgrid Emulator
Fast Gaussian-process emulators of CRK-HACC hydrodynamical summary statistics as a function of subgrid feedback physics.
