Applied AI · Machine Learning · Computer Vision
ASCII-rendered portrait of Klara Katić.
↖ move cursor · press to ripple01 — About
I build models for real-world problems, where performance has to hold beyond the validation split.
I finished a BSc in Computing at the University of Zagreb's Faculty of Electrical Engineering and Computing and stayed on for a master's in Biomedical Engineering — a deliberate move. The interesting problems in machine learning right now aren't in the architecture; they're in what happens when you point a model at clinical data that is messy, imbalanced, and attached to someone's outcome.
02 — Work
Compared U-Net and FC-DenseNet architectures for tumour segmentation on the BraTS 2020 MRI dataset. Built the full pipeline — preprocessing, data augmentation, training, and evaluation against Dice score, precision and recall.
Placed 6th in the AI4Health Croatia Innovation Competition with a model that identifies patients at risk of not taking their medication as prescribed — a problem that quietly undoes a large share of otherwise effective treatment.
An augmented-reality mobile web app that uses GPS geofencing to trigger a 3D Roman guide and contextual audio as you walk through Diocletian's Palace in Split. Third place among more than fifty teams.
Ranked 18th of 75 teams with a Python WebSocket bot combining cross-exchange arbitrage, ETF fair-value estimation, and latency-aware IOC execution — the whole thing decided by microseconds.
Analysed clinical patient data and built predictive models for stroke risk, using statistical analysis to test how demographic, lifestyle and medical variables actually relate to stroke occurrence.
Co-founded a nonprofit initiative connecting citizens, volunteers, businesses and organisations with the people and communities that need them.
A cosmetics storefront with product listings across ten categories, a persistent shopping cart, and basic checkout — server-rendered with Express and EJS.
A note-taking progressive web app built to work with no connection at all — service worker plus application cache, installable to the home screen.
03 — Experience
04 — Education
05 — Stack
06 — Contact
Let's build
something useful.
Open to roles and collaborations in applied AI — research, product, or anywhere the two meet.