Founder, MarkerLeap · Tel Aviv, Israel
Michael Chertok has spent twenty years building AI systems that read the real world, from visual search at scale to consumer-health devices to live-sports prediction. He was a Principal Scientist at Amazon for nine years, first at Lab126 (consumer-health and smart-home devices, including home-camera AI with continual self-learning across customer devices, and physiological sensing such as Amazon Halo), then building the Live Sports AI team at Prime Video (real-time predictive models for NBA and NFL). Earlier he co-founded JustVisual, where as CTO he scaled a real-time visual-search platform to 90M monthly active users and $40M ARR, and was an early force in Israel's deep-learning ecosystem. He began in digital signal processing (Motorola, CEVA) and computer vision for real-time video (Vigilant Technology).
He now runs MarkerLeap, a biomarker discovery lab that recovers biomarkers from the body's hardest-to-read raw signals, and advises AION Labs, the AI-for-drug-discovery venture studio AWS founded with AstraZeneca, Merck KGaA, Pfizer, and Teva, on AI for drug discovery and clinical trials.
In Defense of OCTA: The Reconstruction-Utility Gap in OCT-to-OCTA Synthesis.
OMIA 2026, the Ophthalmic Medical Image Analysis workshop at MICCAI.
arxiv.org/abs/2608.15626
CourtMotion: Learning Event-Driven Motion Representations from Skeletal Data for Basketball.
With Omer Sela and Lior Wolf, 2025.
arxiv.org/abs/2512.01478
Generating Factually Consistent Sport Highlights Narrations.
With Noah Sarfati, Ido Yerushalmy and Yosi Keller. MMSports 2023, the 6th International Workshop on
Multimedia Content Analysis in Sports, at ACM Multimedia.
doi.org/10.1145/3606038.3616157
Balancing Specialization, Generalization, and Compression for Detection and Tracking.
With Dotan Kaufman, Koby Bibas, Eran Borenstein and Tal Hassner. BMVC 2019, the British Machine Vision
Conference.
arxiv.org/abs/1909.11348