CV

PhD, Doctoral Student

Tandon School of Engineering, New York University, New York

• Department of Mechanical and Aerospace Engineering. GPA: 3.9/4
• Advisor: Prof. Katepalli R. Sreenivasan

M.Tech (Research)

Department of Computational and Data Sciences, Indian Institute of Science (IISc) Bangalore, India

• Computational and Data Sciences. GPA: 9.3/10

B.Tech

Department of Mechanical Engineering, Indian Institute of Technology (BHU) Varanasi, India

• Mechanical Engineering. GPA: 8.0/10

PhD Thesis — Decaying Turbulence via Direct Numerical Simulations

New York University

• Ran record-duration pseudo-spectral DNS (to 4096³) over ~200,000 eddy-turnover times, Re_λ = 30–145, resolving the asymptotic decay exponent by separating Birkhoff–Saffman (k²) and Loitsianskii–Kolmogorov–Batchelor (k⁴) regimes; benchmarked against Migdal's field-theoretic predictions.
• Designed a dynamic regridding scheme (coarsening as scales grow) that cut computational cost >5× while preserving turbulence statistics.
• Built a modular single-GPU pseudo-spectral Navier–Stokes solver in C++/CUDA (cuFFT, RK2, divergence-free projection); extending it to a multi-node multi-GPU solver with cuFFTMp/NVSHMEM, validated to machine precision.
• Developing ML surrogates coupling energy-spectrum forecasting with conditional velocity-field generation, validated against DNS.
• Ran production campaigns on TACC (ACCESS), NYU Greene, and KAUST Shaheen III; released reproducibility data and figure scripts.

Research Assistant — Numerical Solver Accuracy Analysis

IIT Kanpur

• Compared pseudo-spectral (Py-Tarang) and finite-difference (Py-Saras) solvers on forced homogeneous turbulence (256³, Re up to ~2000), showing that despite higher per-timestep error, the FD solver matches spectral accuracy in energy evolution, spectra, flux, and velocity-derivative PDFs.
• Argued and demonstrated that numerical errors largely cancel within the turbulence attractor, supporting the use of more scalable FD solvers for large-grid DNS.
• Ran GPU-accelerated 3D simulations using in-house CuPy frameworks on NVIDIA hardware to quantify discretization and spectral accuracy.

Masters Thesis — Hydrogen/Methane Combustion & Dimensionality Reduction

Indian Institute of Science (IISc)

• Performed detailed-chemistry DNS of high-pressure hydrogen/methane wrinkled laminar flames at reheat (sequential gas-turbine) conditions using the massively parallel S3D solver with CHEMKIN/TRANSPORT and reduced San Diego mechanisms.
• Quantified the split between spontaneous-ignition and flame-propagation fuel consumption versus pressure and hydrogen fraction, using chemical explosive mode analysis (CEMA).
• Co-developed co-kurtosis PCA (CoK-PCA), a high-order-moment dimensionality-reduction method that identifies stiff, extreme-valued chemical dynamics (e.g. ignition kernels) more accurately than standard PCA.

Burgers Program Summer School on Turbulence

University of Maryland

ICISS: Near-Wall Reactive Flows

TU Darmstadt (Virtual)

High-Performance Computing Workshop

SERC, IISc Bangalore