Skip to content
View kumarDeepak-su's full-sized avatar
💭
I scan earth for a living.
💭
I scan earth for a living.

Block or report kumarDeepak-su

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
kumarDeepak-su/README.md

Dr. Deepak Kumar

Assistant Professor · Institute of Geophysics, Polish Academy of Sciences
Seismic imaging · Full-waveform inversion · Physics-informed machine learning

LinkedIn ORCID Email


About

I image the Earth's interior with seismic waves, and I build the computational methods that make those images trustworthy.

My work sits between classical geophysics and machine learning: wide-angle and 3-component seismic imaging of the crust and upper mantle, full-waveform and impedance inversion, and deep learning that is constrained by physics rather than replacing it. A recurring theme in my research is calibrated uncertainty — an inversion result is only useful if you know how much to trust it.

  • Lithospheric-scale seismic imaging, crust–mantle structure and seismic anisotropy
  • Full-waveform inversion, velocity model building, tomography
  • Physics-informed deep learning for inverse problems
  • CCUS: seismic characterisation and monitoring for CO2 storage

Previously at CSIR-NGRI (India), where my PhD produced high-resolution crustal velocity and anisotropy models of the Archean Dharwar Craton from 3-C wide-angle seismic data.


Featured research

Project What it does
DEEP_FWI_GIPE Reliability-calibrated deep residual FWI. A geometry-invariant physics encoding lets one trained ensemble transfer to acquisition geometries it has never seen, and a held-out-shot physics audit repairs its confidence intervals without ground truth.
Physics-Informed-Neural-Network ADMM-guided physics-informed deep learning for 2-D acoustic impedance inversion with reweighted ℓ1 sparse regularization.
Seismic_Processing_Inversion_Modelling_WaveSimulation Notebooks for seismic processing, wave simulation and inversion workflows.
Milestone_Project_AIML Applied machine-learning and deep-learning projects: time-series forecasting, CNN image classification, NLP and unsupervised clustering.

Everything is MIT-licensed and reproducible — pre-trained checkpoints, example data and quick-start scripts included.


Selected publications

  • Reliability-calibrated deep residual full-waveform inversion using geometry-invariant physics encodingsubmitted to Computers & Geosciences · arXiv:2607.28535
  • Calibrated uncertainty for wide-angle crustal models: how firmly is the Dharwar Craton Moho actually constrained?submitted to Geophysical Journal International · arXiv:2607.28199
  • ADMM-guided physics-informed deep learning for 2-D acoustic impedance inversionsubmitted to IEEE TGRS · EarthArXiv
  • Deep crustal structure and compositions of the Dharwar Craton from 3-C wide-angle seismic dataJournal of Asian Earth Sciences (2022)
  • Upper-crustal structure along the Perur–Chikmagalur 3-C profile, Archean Dharwar ProvinceActa Geophysica (2023)

Toolbox

Python PyTorch NumPy SciPy MATLAB LaTeX

Seismic software — ProMax · FOCUS (Echoes) · Seismic Unix · Petrel · OpendTect · DSG Landmark · Rayinvr · Deepwave


Open to collaboration on seismic imaging, inverse problems, uncertainty quantification and CCUS monitoring.

Popular repositories Loading

  1. Physics-Informed-Neural-Network Physics-Informed-Neural-Network Public

    ADMM-guided physics-informed deep learning for 2D acoustic impedance inversion with reweighted L1 sparse regularization

    Python 3 1

  2. Milestone_Project_AIML Milestone_Project_AIML Public

    Applied machine-learning and deep-learning projects: time-series forecasting, CNN image classification, NLP and unsupervised clustering

    Jupyter Notebook 1

  3. Seismic_Processing_Inversion_Modelling_WaveSImmulation Seismic_Processing_Inversion_Modelling_WaveSImmulation Public

    Notebooks for seismic data processing, wave-equation simulation, modelling and inversion workflows

    Jupyter Notebook

  4. Seismology Seismology Public

    Seismology-MCS data reading. processing

  5. DEEP_FWI_GIPE DEEP_FWI_GIPE Public

    Reliability-calibrated deep residual FWI with geometry-invariant physics encoding

    Python

  6. kumarDeepak-su kumarDeepak-su Public

    Profile README