Sagar Pal
Sagar Pal

Sagar Pal

I work at the intersection of scientific computing, AI for science and high-performance ML systems.

View projectsDownload CV
Gray–Scott reaction–diffusion: coral-like colonies with teal rims and amber cores spreading across the fieldGray–Scott reaction–diffusion: coral-like colonies with teal rims and amber cores spreading across the field
Gray–Scott reaction–diffusion, solved live on your GPU. Drag across the field to seed new growth.Gray–Scott reaction–diffusion, live. Touch the field to seed growth.
Sagar Pal, outdoors in a park in winter, wearing a dark coat and glasses

From droplets to materials

I started in mechanical engineering, moved into theoretical fluid dynamics, and spent my PhD at Sorbonne Université building numerical methods for multiphase flows: air meeting water at extreme speeds, breaking into droplets and sprays.

A post-doc pushed me into statistical physics, coupling direct numerical simulation with Monte Carlo to model how liquids fragment. Today I lead the technical side of Entalpic's ML platform for materials discovery, where generative and predictive models run next to HPC chemistry.

Across the whole stack, from hand-tuned Fortran on 64,000 cores to PyTorch and JAX on GPU clusters.

Mechanical engineering, then theoretical fluid dynamics, then a PhD at Sorbonne Université on numerical methods for multiphase flows. A post-doc in statistical physics. Now the technical side of Entalpic's ML platform for materials discovery.

  • 2025 – nowSenior ML & HPC Engineer, Entalpic
  • 2021 – 2025Research Engineer, Saint-Gobain Research
  • 2020 – 2021Post-doctoral Researcher, Sorbonne Université
  • 2017 – 2020PhD in Mechanics, Sorbonne Université
  • 2016 – 2017MSc in Fluid Mechanics, UPMC & École Polytechnique
  • 2015 – 2016Research Fellow, Indian Institute of Science
  • 2011 – 2015B.Tech in Mechanical Engineering, Indian Institute of Technology
Full CV

Simulations and tools

Side projects, mostly physics running in a browser tab.

FlowLab: dye shed from a cylinder curling into a Kármán vortex street

FlowLab

Real-time 2D Navier–Stokes in the browser: pressure projection, semi-Lagrangian advection, obstacles you can drag through the flow.

WebGPU, WGSL, Navier–Stokes
Live demoSource
Gray-Scott simulator: spots and loops in amber and teal on a dark field

Gray-Scott

Turing patterns on your GPU: twelve presets, brush seeding and live parameter tuning with vim-style keys.

WebGPU, WGSL, reaction–diffusion
Live demoSource
$ ports

 PORT   PROCESS  PROJECT     FRAMEWORK
 :3000  next     frontend    Next.js
 :5173  vite     preview     Vite
 :5432  docker   postgres    PostgreSQL
 :8000  uvicorn  api         FastAPI

 4 ports active

PortWhisperer

See what's running on your ports, with framework detection, Docker containers and cleanup of orphaned processes.

Python, CLI
PyPISource
Rent vs Buy simulator: input sliders, the verdict, and a ten-year net-value chart where buying overtakes renting after seven years

Rent vs Buy

Compare realistic renting and buying scenarios with mortgage rates, appreciation, returns, taxes and upkeep.

Python, simulation, finance
Live demoSource

Papers

Google Scholar
  1. 2024
    Statistics of drops generated from ensembles of randomly corrugated ligamentsS. Pal, C. Pairetti, M. Crialesi-Esposito, D. Fuster, S. Zaleski. Physics of Fluids.
    PDFDOI
  2. 2021
    A novel momentum-conserving, mass-momentum consistent method for interfacial flows involving large density contrastsS. Pal, D. Fuster, S. Zaleski. Journal of Computational Physics.
    PDFarXiv
  3. 2021
    A mass-momentum consistent, Volume-of-Fluid method for incompressible flow on staggered gridsT. Arrufat, M. Crialesi-Esposito, D. Fuster, Y. Ling, L. Malan, S. Pal, R. Scardovelli, G. Tryggvason, S. Zaleski. Computers and Fluids.
    PDFDOI
  4. 2021
    PArallel, Robust, Interface Simulator (PARIS)W. Aniszewski, T. Arrufat, M. Crialesi-Esposito, S. Dabiri, D. Fuster, Y. Ling, J. Lu, L. Malan, S. Pal, R. Scardovelli, G. Tryggvason. Computer Physics Communications.
    DOI