Mariano聽Tepper
Senior Principal Engineer at Elastic
Portland, Oregon, United States
Summary
馃ぉ
Rockstar馃帗
Top SchoolMariano Tepper is a Senior Principal Engineer based in Portland with 11 years of experience bridging ML research and production-grade systems, currently focused on vector search and similarity algorithms. He combines deep theoretical work鈥攕panning new algorithms and randomized linear algebra鈥攚ith practical, high-performance implementations deployed in industry R&D and product teams. His trajectory includes AI research leadership at Intel Labs and DataStax and recent enterprise engineering at Elastic, reflecting a rare blend of research rigor and systems delivery. An early contributor to scalable linear algebra in the dask ecosystem (TSQR and compressed SVD for tall-and-skinny matrices), he brings hands-on open-source experience optimizing distributed numerical methods. He holds a PhD in Computer Science and a strong applied math background from ENS Cachan, which informs his focus on provable, efficient ML infrastructure.
11 years of coding experience
14 years of employment as a software developer
Master of Science (M.Sc.) Mathematics Vision and Learning, Master of Science (M.Sc.) Mathematics Vision and Learning at Ecole Normale Sup茅rieure de CachanDoctor of Philosophy (Ph.D.) Computer Science, Doctor of Philosophy (Ph.D.) Computer Science at University of Buenos Aires
English, Spanish, French
Github Skills (13)
dask10
python10
svd10
linear-algebra10
numpy10
data-science9
data-structure9
algorithm9
data-structures9
algorithms9
scipy8
pandas8
testing7
Programming languages (3)
Github contributions (5)
dask/dask
Feb 2015 - Sep 2017
Parallel computing with task scheduling
Role in this project:
Data Scientist
Data ScientistContributions:3 PRs, 22 comments, 2 issues in 2 years 7 months
Contributions summary:Mariano primarily contributed to the development of a linear algebra module within the Dask library, focusing on QR and SVD decompositions for tall-and-skinny matrices. Their work involved implementing the TSQR algorithm, adding support for irregular block shapes, and subsequently creating tests to verify the correctness of the QR and SVD implementations. They also introduced a compressed SVD implementation using random sampling techniques.
pythonschedulingparallelnumpydask
marianotepper/marianotepper.github.io
Nov 2015 - Jan 2019
Contributions:151 pushes in 3 years 2 months
Find and Hire Top DevelopersWe鈥檝e analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.