Computational Physicist · PhD

Physics-aware machine learning for atomistic simulation

I build machine-learning surrogates for atomistic simulation, and I handle the whole loop: DFT data, model training, validation against physics and experiment, and deployment on multi-node GPUs.

Bengaluru, India Publishes as A. Shankar Open to computational materials / scientific ML roles
Download CV Email Google Scholar LinkedIn GitHub

About

I'm a computational physicist with a PhD in computational materials science. I have spent six years running molecular dynamics and density functional theory on liquid and glassy metals, and I have six peer-reviewed papers, including a Journal of Applied Physics Featured Article on Mg-Ca metallic glasses as biodegradable bone-graft materials.

Since 2026 I have been building an end-to-end scientific-AI workflow. I fine-tune the MACE-MPA-0 atomistic foundation model on a DFT dataset built for the job, deploy it in LAMMPS on multi-node A100 GPUs, and check it against classical potentials and experiment.

6peer-reviewed papers
6 yrsMD and DFT on liquid and glassy metals
96%weak-scaling efficiency on 8 A100 nodes
3.3×throughput from GPU memory tuning

Current project

Fine-tuning the MACE foundation model for Mg-Ca metallic glasses

Independent Researcher · Co-I, PARAM Rudra GPU grant · 2026 to present

  • Training data. I built a 2,000-structure DFT training set (Quantum ESPRESSO) spanning crystalline, liquid, supercooled and glassy Mg-Ca, with an active-learning round aimed at the model's errors.
  • DFT workflow. I designed the DFT convergence protocol and an automated SLURM workflow for hundreds of calculations, saving about 1,100 CPU node-hours.
  • GPU deployment. MACE runs in LAMMPS on multi-node A100 GPUs (libtorch, Kokkos, CUDA), validated against ASE. It reaches 96% weak-scaling efficiency on 8 nodes, and memory tuning gave up to 3.3× throughput.
  • Profiling and debugging. I profiled with Nsight Systems and traced a multi-GPU crash in the LAMMPS Kokkos ML-IAP interface to its root cause, documented for upstream.
  • Physics validation. Zero-shot MACE captures glass local structure that the classical 2NN-MEAM potential misses: a Mg/Ca icosahedral-centre ratio of 3.3 against 1.0, with experiment at 5.6.
  • Next. Completion is expected in November 2026, with a public release of the fine-tuned MLIP and a journal publication to follow.

Research highlights

Metallic glasses as biodegradable bone grafts

MD of Mg2Ca, Al2Ca and Al4Ca glasses finds glass-transition temperatures of 550 to 689 K, and mechanical and thermophysical properties better than their crystalline counterparts. JAP Featured Article, 2024.

How atoms move in a glass

Mass transport in binary BMGs, explained with the Darken-Manning equation and free-volume theory using MD inputs, plus evidence of Stokes-Einstein breakdown in the supercooled liquid. J. Phys.: Condens. Matter, 2025.

Liquid refractory metals

Structure, diffusion, viscosity and thermal properties of liquid tungsten and molybdenum from the melting point to the boiling point (2900 to 5800 K). Physica Scripta, 2023.

Open source

Partial radial distribution functions of Mg2Ca at 300 K (glass) and 1200 K (liquid). The split second peak in the glass washes out in the liquid.
Mg2Ca partial RDFs. The split second peak at 300 K is the structural signature of the glass, and it is gone by 1200 K.

anjali-shankar/metallic-glass-md

LAMMPS inputs and Python analysis behind the JAP 2024 and JPCM 2025 papers

It contains the published input decks (glass-transition quench, Green-Kubo viscosity, VACF/MSD, tensile), plus one small command-line script per physical quantity: partial S(q), Green-Kubo versus Einstein diffusion, and vibrational density of states. Sample Mg2Ca data at 13 temperatures is included, so every script runs out of the box.

LAMMPSPythonMolecular dynamicsMetallic glass

Experience

Jun 2026 to present

Research Quality Specialist / Subject Matter Expert, Physics and Materials

Turing · Remote · 3 projects

I write and review paper-derived scientific coding tasks for frontier-lab clients. Topics include electrochemistry, Poisson-Boltzmann electrostatics, thermochemical equilibrium, rovibrational spectroscopy and solid-state chemistry. Each task has a reference solution, a rubric and a mutation-tested hidden test suite. I also build agentic LAMMPS tool-use tasks.

Sep 2026 to present

AI Trainer, Scientific Computing

Aligned · Remote

I build containerised terminal-bench style tasks, each with a reference solution and a hidden verifier. They test data triage and choice of method on realistic simulation outputs.

Sep 2026 to present

AI Trainer, Physics

Handshake AI · Remote

I write adversarial multimodal physics tasks (a lab image plus an expert prompt). Current vision-language models fail them for a specific, nameable physical reason.

Feb 2020 to Jan 2026

Doctoral Researcher, Computational Materials Science

Department of Physics, MK Bhavnagar University · Supervisor: Prof. N. K. Bhatt

I ran large-scale LAMMPS molecular dynamics on Mg2Ca, Al2Ca, Al4Ca and Ca61Al39 (10,000 atoms, 300 to 1500 K, 2NN-MEAM), and on liquid W and Mo from melting to boiling point (2900 to 5800 K). I extracted diffusion, viscosity, glass-transition temperatures, elastic and thermodynamic properties and structure (RDF, S(q), CNA), and showed Stokes-Einstein breakdown in supercooled metallic glasses. I cross-checked the results against Quantum ESPRESSO calculations and built reusable Python analysis pipelines (metallic-glass-md).

2015 to 2024

Visiting Faculty and Lecturer, Physics

MK Bhavnagar University and affiliated colleges

I taught computational and condensed matter physics at MSc level, supervised LAMMPS-based MD dissertations, and ran workshops on MD and visualisation.

Publications

Published under the author name A. Shankar. Google Scholar profile.

  1. Atomic mass transport in binary metallic glasses: a case study of Mg2Ca, Al2Ca and Al4Ca
    A. Shankar, N. K. Bhatt · Journal of Physics: Condensed Matter 37, 355401 (2025)
  2. Enhanced mechanical and thermophysical properties of Mg2Ca, Al2Ca, and Al4Ca bulk metallic glasses in comparison to crystalline alloys for bone grafting applicationsFeatured
    A. Shankar, N. K. Bhatt · Journal of Applied Physics 136(3), 035107 (2024)
  3. Bulk metallic glass Al2Ca from a biomedical application viewpoint: a molecular dynamics study
    A. Shankar, N. K. Bhatt · IOP Conf. Series: Materials Science and Engineering 1291, 012026 (2023)
  4. Mass transport and thermal properties of liquid (melting to boiling point) tungsten: a molecular dynamics simulation
    D. R. Gohil, A. Shankar, N. K. Bhatt · Physica Scripta 98, 115963 (2023)
  5. Molecular dynamics simulations for liquid molybdenum: structural and atomic transport properties
    D. R. Gohil, A. Shankar, N. K. Bhatt · Bulgarian Journal of Physics 50(4) (2023)
  6. Amorphous structure of binary Ca61Al39 metallic glass close to transition temperature
    A. Shankar, N. K. Bhatt, A. B. Patel, P. R. Vyas · AIP Conference Proceedings 2352, 020038 (2021)

Skills

Scientific machine learning

MACE-MPA-0 fine-tuningML interatomic potentialsDFT training-set designActive learningValidation vs experimentScaling benchmarks

Physics-based simulation

LAMMPS2NN-MEAM / EAMStillinger-WeberTight bindingQuantum ESPRESSOASE

HPC and deployment

SLURMMulti-node A100Kokkos + CUDA 12.4libtorchNsight SystemsNVIDIA NGCOpenMPI

Programming

PythonNumPy / SciPy / PandasPyTorchBashGit

Modeling and analysis

Statistical mechanicsLiquid-state theoryGreen-Kubo transportStokes-EinsteinRDF / S(q) / CNA / VoronoiTime correlation functions

AI training and evaluation

Scientific coding tasksAgentic terminal tasksMultimodal physicsRubricsHidden test suites

Education

PhD, Physics (Computational Materials Science)
MK Bhavnagar University, Gujarat · 2020 to 2026
MSc, Physics (Condensed Matter)
Sardar Patel University, Gujarat · 2013 to 2015
BSc, Physics (ranked 3rd in cohort)
Sir P. P. Institute of Science, Gujarat · 2010 to 2013

Recognition

PARAM Rudra A100 allocation (NSM)
Co-Investigator · 2026
1st Prize, Oral Presentation
National Conference at IITE, Gandhinagar · 2022
2nd Prize, Poster Presentation
National Science Symposium, Govt. Science College, Gariyadhar · 2024
Workshops and schools
ICTP Young Researchers' Workshop on ML for Materials (2022); ICTP Total Energy and Force Methods (2021); IISc Creep and Fracture conference (2024); IIT Kharagpur MD and glass-science courses (2020, 2021); 9 conference presentations

Contact

I'm open to computational materials scientist and scientific ML roles, particularly in Bengaluru. The quickest way to reach me is anjalirameja1993@gmail.com.

Email me Download CV (PDF)