Senior Libraries Engineer · NVIDIA

Aydın Buluç

Parallel algorithms.
Sparse computation.
Scientific discovery.

I work at the intersection of high-performance computing, sparse linear algebra, and combinatorial algorithms. I care about algorithms that are mathematically sound, scalable in practice, and useful on real systems.

At NVIDIA, I primarily work on our next-generation iterative solvers (cuISS), but I also frequently collaborate with cuDSS and cuSPARSE teams.

Research & software

Explore the work

01 / ALGORITHMS

Sparse linear algebra

Parallel sparse matrix and tensor computations, communication-avoiding methods, and GPU-accelerated solvers.

02 / SOFTWARE & STANDARDS

Graphs through linear algebra

GraphBLAS C API and standards leadership, Combinatorial BLAS, and graph computation on GPUs.

03 / SCIENTIFIC APPLICATIONS

Computational biology

HipMCL and PASTIS: scalable clustering and sequence similarity for discovering protein families.

Background

Before joining NVIDIA in November 2025, I spent over fifteen years at Lawrence Berkeley National Laboratory. I led PASSION Lab, a research team of UC Berkeley Ph.D. students and LBNL postdoctoral fellows, research scientists, and engineers.

I was the director of Sparsitute from its founding in September 2022 until November 2025. The DOE mathematical center spans six institutions and more than 30 members. I also served on the UC Berkeley EECS adjunct faculty from 2017 to 2025.

I received my Ph.D. in Computer Science from UC Santa Barbara in 2010, advised by John R. Gilbert, and my B.S. from Sabancı University in 2005.

Teaching & mentoring

Selected recognition

  • 2025
    ACM Distinguished Member

    Contributions to high-performance algorithms and software.

  • 2022
    ACM Gordon Bell Prize finalist

    Team leader, extreme-scale protein similarity search.

  • 2015
    IEEE TCSC Early Career Award

    Excellence in scalable computing.

  • 2013
    DOE Early Career Award

    Office of Science.