2025 · Active
AutoPerf — HPC benchmarking & optimization platform
Agent-driven benchmarking and optimization workflows for CUDA, OpenMP, and MPI stacks.
PerfHPCAI
Projects
Instruments for HPC, performance, and visual systems.
Explore by domain, tech stack, or performance characteristics. Details slide in without breaking context.
2024 · Joint with Numpex/ONERA
Samurai solver modernisation
Porting research CFD kernels to a maintainable release track.
HPCCUDA
2023 · Research
Distributed sparse neural networks
MPI-enabled sparse models tuned for exascale targets.
HPCAI
2022 · Lab
CFD neural operators
Neural surrogates predicting pressure and velocity fields.
CFDAI