I think about black holes for a living: the BPS kind, mostly, and what their spectra can tell us about Donaldson–Thomas invariants.
Defended my PhD at Sorbonne Université / LPTHE in June 2026, supervised by Boris Pioline. Since then I've been retooling toward scientific computing: finished the CERN STEAM Academy 2026, a 10-week program covering heterogeneous CPU/GPU computing, FPGA edge inference, and applied deep learning. The through-line is the same either way: build something rigorous, make it fast, check it against reality.
- Benchmarking neural discriminators (transformers, PFNs, XGBoost) for an LHC smuon search, with LPTHE/LPNHE
- Looking for what's next in accelerated scientific computing / ML for physics
- Jejjala, Mondkar, Mukhopadhyay & Raj, Learning Holographic Horizons, Phys. Rev. D 111, 026016 · arXiv:2312.08442
- Pioline & Raj, Black Hole Quantum Mechanics and Generalized Error Functions, JHEP 03, 179 · arXiv:2507.08551
- Le Floch, Pioline & Raj, BPS Dendroscopy on Local P¹ × P¹, Annales Henri Poincaré · arXiv:2412.07680
|
matrix-models-cpp
|
r-distro
|
|
xilinx-counter
|
OpenMP_Scaling_Benchmark
|
C++17/20/23 Python PyTorch CUDA Julia Mathematica SageMath · MPI, PBS, ROOT, Boost, GSL · CMake, Docker, GitHub Actions