Sabri 'Elras' Bolkar

Linkedin | Twitter | Mastodon | Github
Email: sabribolkar (at) gmail (dot) com


I am a machine learning applied scientist, engineer, and basketball player. I am interested in the complete lifecycle of learning-based systems from research to development, deployment, scaling, and improvement. I worked with the full spectrum of data (tabular, graph, language, image, video, 3d), and developed various discriminative or generative models from LLMs to diffusion operating on such data.

I studied computer vision at NTNU and completed my MSc thesis on unsupervised image segmentation at KU Leuven. Then I had a PhD adventure at TU Delft where I worked on fast registration algorithms for incomplete point cloud data.

Currently, I am working on massive scale ML systems at Bol, the largest online shop in Benelux. Previously, I researched deep networks and led architecting the ML cloud infrastructure at one of the largest visual content companies, Bright River. You definitely saw my models in action on your favorite e-commerce website (;

I also write about deep learning at InfoQ, you can check my articles here.


Updates

July 23 - My EU AI Act summary reached 2nd in Hacker News, triggered interesting comments as well.

March 23 - Check Pos3D, our realistic on-model imagery service.

June 22 - We scaled our high resolution image editing ML API usage to millions per month.

July 21 - Our new ML image editing SaaS ModifAI is live!

February 21 - The project that I am leading the R&D efforts got NVIDIA Inception Grant and several cloud grants.

March 20 - Check my team's deep learning solutions as we bring automation to image editing.

April 19 - Lured to the industry from my PhD researcher position. I will be initializing the CV/ML group at B/R.

November 18 - My work on microscopy point cloud registration resulted in state-of-the-art speed with high quality!

July 18 - Defended my thesis succesfully! I will be starting a PhD positon at the TU Delft Quantitative Imaging Group.

September 16 - Started my MSc education, my specializations will be imaging, processing and machine learning.

May 15 - Received the best poster award for my brain-computer interface design at the DTU.

June 14 - Interned at Technical University of Denmark, I worked on brain-computer interfaces and EEG data analysis.


Projects | Publications

Scaling Deep Networks Series
InfoQ, 2021-
Articles link

High Resolution Alpha Masking of Objects
B/R, 2020
Sample input image from link
Another example result for link

Soft Segmentation of Viral Labeled Neurons
Sabri Bolkar
MSc Thesis, KU Leuven (Neuro-electronics Research Flanders) & Norwegian University of Science and Technology, 2018
[Pdf] [Code]

Deep Smoke Removal from Minimally Invasive Surgery Videos
Sabri Bolkar, Concong Wang, Faouzi Alaya Cheikh, Sule Yildirim
IEEE International Conference on Image Processing (ICIP), 2018
[Pdf] [Code]

Bio-spectral Imaging
Sabri Bolkar, Onur Ozcelik
Research Report, METU EE, 2015
[Pdf]