Hameer Abbasi

Timeline

1. 2014–2015

   Engineering studies in Pakistan and Germany

   He completes a bachelor's degree in electrical (telecommunications) engineering at the National University of Sciences and Technology in Pakistan in July 2014, then works at LMK Resources in Pakistan until September 2015. He then moves to Germany for a master's in information and communication engineering at Technische Universität Darmstadt.

2. By 2018

   Maintaining PyData/Sparse

   In a Hilfswissenschaftler job, a research-assistant role in Germany, he meets the problem of scaling a sparse system and discovers PyData/Sparse, then kept in Matthew Rocklin's personal repository. He becomes the project's maintainer, and the work brings him a client in Quansight, where he starts as a contractor.

3. 2018

   SciPy 2018 sparse array talk

   At SciPy 2018, he presents 'Sparse - A More Modern Sparse Array Library' as a speaker from TU Darmstadt: multi-dimensional sparse arrays that follow NumPy's array interface and work with Dask and XArray. His proceedings paper describes a package that builds on scipy.sparse and supports dimensions greater than two.

4. By 2020

   uarray, unumpy and udiff

   In 2020, he works on client projects at Quansight and on projects in its Labs division, among them uarray, a backend-dispatch system, unumpy, a backend-agnostic version of NumPy, and udiff, an automatic differentiation library built on top of unumpy. He has also started new research on PyData/Sparse.

5. 2020

   Coauthoring the NumPy paper

   In September 2020, 'Array programming with NumPy' appears in Nature with him among its authors, alongside Ralf Gommers, Pearu Peterson and Travis E. Oliphant of Quansight. The paper reviews the concepts of NumPy, the primary array programming library for Python.

6. January 2024

   DARPA funding for sparse

   In January 2024, he announces that Quansight has received funding from DARPA, together with partners from MIT, under its Small Business Innovation Research program to build out sparse with just-in-time compilation. The plan adopts the Array API standard as the interface, a standard championed by NumPy, PyTorch and CuPy.

7. June 2025

   Finch and PyData/Sparse at PLDI

   In June 2025, he and Mateusz Sokol present 'PyData/Sparse & Finch: extending sparse computing in the Python ecosystem' in the Sparse 2025 track at PLDI 2025 in Seoul. The talk covers making the Finch sparse tensor compiler and Galley scheduler available to the community and standardizing aspects of sparse computing.

8. As of October 2026

   Software engineer at Quansight

   He is a software engineer at Quansight, based in Luxembourg, and his main area of work is sparse arrays, mostly in Python.