Ibrahim Jubran
Ibrahim Jubran is a computer scientist whose research on coresets, a data-summarization technique, speeds up machine-learning algorithms on large datasets. He earned a B.Sc. in 2014 and an M.Sc. in 2018 at the University of Haifa, and received a PhD under Dan Feldman in the university's Robotics & Big Data Labs. A 2019 NeurIPS paper on least-mean-squares solvers, coauthored with Alaa Maalouf and Feldman, received an Outstanding Paper Honorable Mention, and a 2021 NeurIPS paper on coresets for decision trees was a spotlight. After working as a computer vision researcher at IBM Research in Haifa, he co-founded the augmented-reality startup Orb-AR and served as its CTO from 2022 to 2024. He is Chief Scientist at Genfied, a company that analyzes in-store retail operations through existing CCTV cameras.
Timeline
-
2014
A B.Sc. at Haifa
He earns a B.Sc. with first class honors in 2014 through the Etgar program in the Computer Science Department at the University of Haifa.
-
2018
Master's and PhD with Feldman
He completes an M.Sc. summa cum laude in 2018 in the Computer Science Department at the University of Haifa, advised by Dan Feldman. He holds a PhD in computer science, also under Feldman, from the university's Robotics & Big Data Labs.
-
2019
Least-mean-squares solvers at NeurIPS
In 2019, he coauthors Fast and Accurate Least-Mean-Squares Solvers with Alaa Maalouf and Dan Feldman, and NeurIPS gives it an Outstanding Paper Honorable Mention. The algorithm fuses the data-summarization techniques known as sketches and coresets, and it speeds up existing solvers such as those in scikit-learn by up to 100 times.
-
2020
Sets Clustering at ICML
In 2020, he coauthors Sets Clustering with Murad Tukan, Alaa Maalouf and Dan Feldman at the International Conference on Machine Learning. The paper obtains the first approximation scheme (PTAS) for the sets-k-means problem that runs in near-linear time.
-
2020–2021
Computer vision at IBM Research
He works as a computer vision researcher at IBM Research in Haifa from 2020 to 2021, on deep learning-based image generation and segmentation solutions for longitudinal medical images.
-
2021
Decision trees and point clouds
In 2021, he coauthors Coresets for Decision Trees of Signals, a NeurIPS spotlight paper with Ernesto Evgeniy Sanches Shayda, Ilan Newman and Dan Feldman; its coresets speed up random forests and their parameter tuning by up to 10 times in sklearn and lightGBM experiments. At ICCV 2021, he coauthors Provably Approximated Point Cloud Registration with Alaa Maalouf, Ron Kimmel and Dan Feldman, whose algorithms give the first provable constant factor approximation for point cloud alignment in O(n) expected time.
-
2022–2024
Co-founding Orb-AR
He is co-founder and CTO of Orb-AR from 2022 to 2024. The startup builds a no-code augmented reality platform for mobile that lets marketers create and launch AR experiences, and a team of 3 developers under his lead reaches a stable product with paying customers.
-
Since 2025
Chief Scientist at Genfied
He joins the founding team of Genfied in 2025 and serves as its Chief Scientist, where he has led the computer vision and AI research. The company analyzes in-store retail operations through retailers' existing CCTV cameras, and the team has submitted 2 patents.
Career
- Since 2025GenfiedChief Scientist
- 2022–2024Orb-ARCo-founder and CTO
- 2020–2021IBM ResearchComputer Vision Researcher
Education
- 2018University of HaifaM.Sc.
- 2014University of HaifaB.Sc.
Sources and credits
Photo: We have no published photo of Ibrahim Jubran yet. Send us a free-licensed photo.
- Ibrahim Jubran
- NeurIPS 2019 Awards
- Fast and accurate least-mean-squares solvers - University of Haifa
- Sets Clustering
- [2003.04135] Sets Clustering
- NeurIPS Spotlight Coresets for Decision Trees of Signals
- [2110.03195] Coresets for Decision Trees of Signals
- ICCV 2021 Open Access Repository: Provably Approximated Point Cloud Registration


