May 2025
Sports Analytics Computer Vision System
Master's thesis: a computer vision pipeline that automates two off-the-ball football statistics, body orientation at the pass and scanning checks before receiving.
Computer Vision · YOLOv8 · PyTorch→
▸ profile: sobhaan▸ now: KPMG · Amsterdam
Sobhaan ul Husan
/AI Engineer▸ kpmg · health_platform
01 / 05
Based in Amsterdam. Building a global health research platform and a hospital AI patient-assistant avatar at KPMG. Before that, research at VU Amsterdam on sports analytics with computer vision, and on social robots in healthcare and education. M.Sc. AI, VU Amsterdam · B.Sc. AI, UvA.
▸ feed:pub · ACM MMSports '25 · body orientation in footballpub · ACM/IEEE HRI '26 · personalized reading robot for primary schoolspub · IDC '26 (under review) · social robot for pediatric medical hypnosiscert · AI-102 Azure AI Engineerlangs · en (pro) · nl (native) · ur (working)now · KPMG · Amsterdam · health platform + AI avatarpub · ACM MMSports '25 · body orientation in footballpub · ACM/IEEE HRI '26 · personalized reading robot for primary schoolspub · IDC '26 (under review) · social robot for pediatric medical hypnosiscert · AI-102 Azure AI Engineerlangs · en (pro) · nl (native) · ur (working)now · KPMG · Amsterdam · health platform + AI avatar
▸ showing: recent_work
Selected projects
Feb 2026in_progress
KPMG: Hospital AI Patient Assistant
AI-powered patient-facing avatar for a hospital setting. Answers patient questions to take routine load off nursing staff, built on Microsoft Foundry agents with a React frontend.
AI Engineering · React · Microsoft Foundry→
Jul 2026in_progress
OneTwo: Daily Football Guessing Game
Personal project, live at onetwofootie.com: a daily football puzzle where you name the player from their career path. Two modes, a Wikipedia and Wikidata ingest pipeline, and a difficulty model that keeps the daily round winnable.
Full-Stack · Next.js · Data Pipeline→
Mar 2026in_progress
KPMG: Global Health Research Platform
Full-stack work on a global research platform that lets researchers across multiple countries publish, manage, and collaborate on health research projects.
Full-Stack · Node.js · Playwright→
Sep 2024
Medical Image Segmentation
Deep learning pipeline for 3D thoracic organ segmentation from CT scans. nnU-Net 3D vs a custom U-Net variant vs ENet baseline on the SegTHOR challenge.
Medical Imaging · Deep Learning · PyTorch→
Dec 2024
AKI Prediction in Septic Shock
Predicting Acute Kidney Injury stages in ICU patients with septic shock using Decision Trees, XGBoost, and SHAP explainability on the AmsterdamUMC database.
Machine Learning · XGBoost · SHAP→
▸ about: short
I build AI systems that live in the real world: hospitals, football pitches, classrooms.
My research spans computer vision, human-robot interaction, NLP, and full-stack development. I care about systems that survive contact with real users, the last 10% where research becomes product.
read_more →