Postdoctoral Researcher · LMU Munich

Dr.Sai Vamsikrishna Isukapalli

Computational chemist working at the intersection of ab initio molecular dynamics, quantum chemistry, and machine learning interatomic potentials. Currently at AK Fingerhut, LMU Munich, studying excess proton transport in liquid water and building DFT-quality datasets for ML potentials using the MACE framework.

Google Scholar LinkedIn GitHub Email Download CV

AK Fingerhut
Dept. of Chemistry
LMU Munich

Research interests

Ab initio Molecular Dynamics

Long-timescale AIMD of protonated liquid water on national HPC systems using CP2K, generating large-scale DFT-quality trajectory data for structural and dynamical analysis.

ML Interatomic Potentials

Generation of energy/force datasets for message-passing ML potentials (MACE, PaiNN/SchNetPack). Dataset curation, model validation, and benchmark MD against experiment.

Nonadiabatic Quantum Dynamics

Excited-state PES computation and quantum dynamics (MCTDH, SHARC) for organic chromophores. Quantifying intersystem crossing, internal conversion, and time-resolved spectra.

Scientific Workflow Engineering

Building modular Python analysis pipelines for multi-GB trajectory data — proton-transfer detection, RDFs, coordination metrics, and reproducible HPC workflows.

Software & projects

auto-eln-logger

Automatic Electronic Lab Notebook logger for computational chemistry HPC clusters. Integrates Slurm job submission with eLabFTW by auto-parsing CP2K, Molpro, ORCA, and Amber input/output files. Zero extra effort from researchers — just replace sbatch with bmdsubmit.

python slurm cp2k molpro elabftw hpc

GitHub →

kevalzug

A chess opening preparation tool that discovers ‘Trojan’ deviations — slightly inferior moves that demand rare, precise defence from the opponent. Explains counterintuitive engine moves by identifying upcoming only-move defence points and likely human misses. Interfaces with the Lichess explorer API.

python chess lichess api stockfish opening prep

private

AIMD trajectory analysis toolkit

Modular Python workflows for streaming and analysing multi-GB XYZ trajectory files from AIMD simulations. Extracts proton-sharing coordinates, RDFs, coordination metrics, and proton-transfer events at scale.

python aimd trajectory analysis excess proton

private

MACE training pipeline for aqueous systems

End-to-end pipeline for generating, curating, and validating DFT-quality energy/force datasets (497k frames) for MACE message-passing ML interatomic potentials. Includes dataset splitting, loss monitoring, and benchmark validation MD.

pytorch mace mlip schnetpack tensorboard

private

Dr.Sai Vamsikrishna Isukapalli

Ab initio Molecular Dynamics

ML Interatomic Potentials

Nonadiabatic Quantum Dynamics

Scientific Workflow Engineering

QC & MD codes

ML & data

Programming

Infrastructure