About

Mohammad Dindoost

High-Performance Computing (HPC), Parallel Graph Algorithms, Graph Neural Networks (GNNs), Graph Representation Learning, Knowledge Graphs.

Lead developer of HiPerMotif, HiPerXplorer, and VF2-PS, open-source parallel algorithms for large-scale graph motif discovery.
Co-author of MoMo (collaboration with Harvard, RWTH Aachen, Zuse Institute Berlin, University of Würzburg), applied in neuroscience connectome analysis.
Active contributor to NSF-funded projects on scalable community detection, graph coarsening, and HPC frameworks.
Developing new GNN models: Motif-Augmented Message Passing (MMP) and TempAnom-GNN (temporal fraud detection).
Extending Combinatorial Multigrid (CMG) to graph learning (pooling, unpooling, coarsening).
Contributor to Arkouda-NJIT, extending Chapel-based HPC analytics.

Outstanding Student Paper Award, IEEE HPEC 2025 (Best Paper Candidate).
Invited presentations at SIAM CSE 2025 and IEEE HPEC 2024; scheduled talks at IEEE HPEC 2025 and ChapelCon 2025.
Research outputs cited internationally (11 citations, h-index 2, Google Scholar).
Codes and methods span critical domains: neuroscience, bioinformatics, fraud detection, and HPC infrastructure.