// PRINCIPAL DATA SCIENTIST, NIKE, INC.

BELHAL KARIMI

DATA SCIENTIST

Principal Data Scientist across Nike's core domains, partnering with orgs like Design & Commerce. 15+ papers. Two startup exits.

◈ LATEST UPDATES
MAR '23

Monk AI acquired by ACV Auctions.

FEB '23

Brainattic acquired by Master The Monster.

MAR '22

Joined Nike to lead Search Retrieval & Ranking.

◈ THE STACK I REACH FOR

Generative AI & Deep Learning

FLUX / SDXL / SD3 3D AI Federated Learning Compression Algorithms

Advisory

Computer Vision Generative Video Healthcare AI Technical Due Diligence

Search (Match & Rank)

Multimodal Retrieval Neural Ranking RAG Embedding Modeling
◈ CURRENT FOCUS

AI FOR
DESIGNERS

Principal Data Scientist, Nike. Building generative tools that move design faster, not just differently.

01 · COLORWAY ENGINE

One franchise,
infinite colorways

Generative models that auto-produce on-brand colorway options for Nike's top franchises, turning a single silhouette into dozens of production-ready palettes in minutes, not weeks.

02 · SKETCH TO 3D

From napkin sketch
to 3D shoe

Custom post-trained generative models take a designer's rough sketch and generate a full 3D model of the shoe, collapsing the idea-to-prototype loop from weeks to hours.

IDEASKETCHPOST-TRAINED MODEL3D SHOE
◈ SPONSORSHIP DECK

ADVISORY
DEALS

Signed as scientific advisor. Some deals close with a trophy.

Monk AI

EXITED

Scientific Advisor · Paris, FR

Damage-detection computer vision built on Mask R-CNN. Acquired by ACV Auctions.

VISIT ↗

Brainattic

EXITED

Scientific Advisor · Montreal, CA / Paris, FR

Search engine in video streams & automated trailer generation. Acquired by Master The Monster.

VISIT ↗

EyeCareX

ACTIVE

Scientific Advisor · Vancouver, BC, CA

AI-enabled at-home eye tests connected to remote optometrists.

VISIT ↗
◈ TROPHY CASE

RESEARCH
DROPS

Peer-reviewed plays. Drag the shelf to scroll.

NEURIPS · 2019

On the Global Convergence of (Fast) Incremental EM Methods

NEURIPS · 2020

Towards Better Generalization of Adaptive Gradient Methods

COLT · 2019

Non-asymptotic Analysis of Biased Stochastic Approximation

ICLR · 2022

Distributed Adaptive Optimization with Gradient Compression

PH.D. THESIS · 2019

Nonconvex Optimization for Latent Data Models

ALT · 2022

Minimization by Incremental Stochastic Surrogate

UAI · 2023

Layer-wise & Dimension-wise Locally Adaptive FL