Introduction
Hello, My name is Shabbir Marzban. I’m a seasoned Applied Scientist with a passion for Computer Vision, Machine Learning, Deep Learning, and Geospatial Data Science. Over the past ten years, I’ve spearheaded innovative projects at top organizations, developing advanced AI solutions that range from mapping and brand analytics to medical imaging. Currently, I’m working on algorithms that enable speed limit indications in navigation technology, improving the reliability of digital maps. With a solid academic foundation in Computer Science and Electrical Engineering, I thrive on turning complex challenges into tangible, impactful technology solutions.
Email | Google Scholar | Blog | LinkedIn | WhatsApp
Work Experience
Staff Applied Scientist, TomTom
Dec 2023 – Present
- Working on algorithms that enable speed limit indications in navigation technology, improving the reliability and efficiency of digital maps.
Senior Machine Learning Researcher, Promaton
June 2021 – Nov 2023
- Led research and development of AI-based tooth shape estimation and prediction techniques for dental imaging solutions.
- Enhanced project management and software engineering skills while developing AI-driven solutions.
R&D Engineer, Navinfo Europe
Jan 2019 – May 2021
- Researched and developed RGPNet, a semantic segmentation network for street-level images, with a training technique that reduces energy consumption by 75%.
- Worked on object detection and segmentation models optimized for limited compute resources.
- Developed methods to detect wear on road markings using dashcam video data.
Computer Vision Engineer, Uru (acquired by Adobe)
Jan 2017 – May 2018
- Developed deep learning models for brand analytics and brand safeness scoring in videos.
- Built a detector for 800+ unique brands (e.g., Coca-Cola, Nike) to analyze video content.
- Researched and implemented deep learning models to extract geometry from single-image depth predictions for 3D augmentations.
- Co-inventor on two patents owned by Adobe.
Research Engineer (Computer Vision), Ingrain (ingrain.io)
Oct 2015 – Dec 2016
- Developed mobile Augmented Reality tracking technology beyond Pokémon GO.
- Built algorithms for automatic ad placements in videos, prototyped in Matlab and deployed in C++ (OpenCV).
- Led a team of recent graduates, optimizing tracking technology in non-static videos.
Research Assistant, IUI and MVG Labs, Koç University
Sept 2013 – Aug 2015
- Researched recognition, detection, and synthesis of affective events in speech and gestures.
- Implemented a system to detect affect bursts (laughter, breathing sounds) from multi-modal input.
Research Assistant, Computer Vision Lab, Lahore University of Management Sciences (LUMS)
Sept 2011 – July 2013
- Researched a novel 3D reconstruction approach from multiple camera feeds, published in ICCV.
- Worked on automatic 3D structure recovery of heritage sites using aerial images and videos.
- Conducted tutorials and supervised projects on non-rigid structure recovery.
Teaching Experience
Koç University
Graduate Teaching Assistant
- COMP 408/508: Computer Vision and Pattern Recognition (Sept 2014 – Jan 2015)
- COMP 132: Advanced Programming (Jan 2014 – May 2014)
- COMP 131: Introduction to Programming (Sept 2013 – Jan 2014)
Lahore University of Management Sciences (LUMS)
Undergraduate Teaching Assistant
- CS 436: Computer Vision Fundamentals (Aug 2012 – Dec 2012)
Blog posts
-
Location History
Recently I pulled all the location history Google has collected over the years on me and I was amazed at the scale of it, there are more than ~2M data points! I queried on it using DuckDB and visualized it using KeplerGL
-
Urdu Fonts
Easy fix for broken rendering of urdu across ubuntu and in chrome in general: