👋 Hi, I'm Iason Skylitsis: an MSc Artificial Intelligence student at the University of Amsterdam, currently writing my thesis at QurAI (Amsterdam UMC) on label-efficient 3D segmentation for body composition analysis in CT scans. I'm drawn to causality, medical AI, and the builder side of research: taking models out of notebooks and shipping them as real tools, from production ETA predictors to hackathon products with paying users in 48 hours.
- Master's thesis: few- and zero-shot 3D segmentation of novel tissue classes using foundation model encoders (DINO, TAP-CT) on CT scans
- Causal discovery meets bandits: extending Structural Causal Bandits to equivalence classes (CPDAGs, PAGs)
- canvas-dl: a CLI that downloads and organises Canvas course files in seconds
- Causal inference and causal discovery research
- Medical imaging and AI for healthcare
- Hackathons and early-stage product experiments
- How we built MegaMinds at MEGATHON 🧠: idea to deployed app with 50+ signups and 5 paying users in 48 hours
- Orbit, a voice-first real-estate lead engine built at {Tech: Europe} London
- Causal bandits, MRI reconstruction, Alzheimer's detection with explainable AI
- Studying AI at UvA, or moving from software engineering into ML research
