Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I'm Matej Marić, a machine learning engineer based in Zagreb, Croatia. I build end-to-end CV/ML systems, from data pipelines and model training to production inference, with a focus on reliable, scalable solutions for real-world footage and imagery. Most recently I worked on a photography-detection system for a museum and on backend ML services for a retail app. I enjoy turning complex data problems into maintainable software, collaborating with cross-functional teams, and applying ML to wildlife monitoring and conservation as well as digital experiences.…I'm Matej Marić, a machine learning engineer based in Zagreb, Croatia. I build end-to-end CV/ML systems, from data pipelines and model training to production inference, with a focus on reliable, scalable solutions for real-world footage and imagery. Most recently I worked on a photography-detection system for a museum and on backend ML services for a retail app. I enjoy turning complex data problems into maintainable software, collaborating with cross-functional teams, and applying ML to wildlife monitoring and conservation as well as digital experiences.

Matej Marić

Data Scientist, AI Engineer, AI Developer, +2





I'm Matej Marić, a machine learning engineer based in Zagreb, Croatia. I build end-to-end CV/ML systems, from data pipelines and model training to production inference, with a focus on reliable, scalable solutions for real-world footage and imagery. Most recently I worked on a photography-detection system for a museum and on backend ML services for a retail app. I enjoy turning complex data problems into maintainable software, collaborating with cross-functional teams, and applying ML to wildlife monitoring and conservation as well as digital experiences.…I'm Matej Marić, a machine learning engineer based in Zagreb, Croatia. I build end-to-end CV/ML systems, from data pipelines and model training to production inference, with a focus on reliable, scalable solutions for real-world footage and imagery. Most recently I worked on a photography-detection system for a museum and on backend ML services for a retail app. I enjoy turning complex data problems into maintainable software, collaborating with cross-functional teams, and applying ML to wildlife monitoring and conservation as well as digital experiences.

Available to hire

I’m Matej Marić, a machine learning engineer based in Zagreb, Croatia. I build end-to-end CV/ML systems, from data pipelines and model training to production inference, with a focus on reliable, scalable solutions for real-world footage and imagery. Most recently I worked on a photography-detection system for a museum and on backend ML services for a retail app.

I enjoy turning complex data problems into maintainable software, collaborating with cross-functional teams, and applying ML to wildlife monitoring and conservation as well as digital experiences.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Language

Croatian

Fluent

English

Fluent

Chinese

Intermediate

German

Beginner

French

Beginner

Work Experience

Machine Learning Engineer at Museum of Hangovers

January 1, 2025 - March 1, 2026

Built a photography-detection system for a 16-camera museum setup, combining YOLO person/phone detection, pose estimation, ByteTrack, and homography mapping to place events on a shared floorplan, plus other CV-based solutions for camera footage analysis. Trained and tuned the event-classification stack: pose-feature extraction, XGBoost pose classifier, plus temporal voting and orientation logic to make detections usable on real footage.

Machine Learning Engineer at Megatrend d.o.o.

July 1, 2022 - November 1, 2024

Built and maintained backend ML services for a retail app where users capture store-shelf photos for item detection/classification. Improved model performance through dataset curation, annotation refinement, retraining, and systematic benchmark/evaluation comparisons. Researched and prototyped experimental features, then integrated selected improvements into production backend workflows.