Jeronim Matijević
Hi, I’m Jeronim Matijević, a machine learning engineer based in Split, Croatia, with around five years of experience building ML, research, and software solutions.
About
Hi, I’m Jeronim Matijević, a machine learning engineer based in Split, Croatia, with around five years of experience building ML, research, and software solutions. I have a strong affinity for computer vision, but I’m passionate about neural networks for a wide range of problems and applications, from perception to optimization. I enjoy turning ideas into robust, production-ready systems and thrive when collaborating across ML research, data, and engineering to deliver scalable solutions.
I’ve shipped end-to-end ML/AI projects on cloud platforms, with a focus on deployment, scalability, and maintainability. My strengths include fast prototyping, designing practical evaluation metrics, and translating complex research into reliable production pipelines using modern MLOps practices (AWS, Kubernetes, Docker, CI/CD). I’m always eager to learn, share knowledge, and tackle new challenges across domains.
- 5 years' experience
- Joined March 2026
Languages
- EnglishFluent
- CroatianFluent
Industry experience
- Computers & Electronics
- Software & Internet
- Media & Entertainment
- Retail
Work experience· 4 roles
Developed a user ID verification service using a hybrid of visual-language models and traditional CNNs; implemented a planogram optimisation service leveraging hierarchical Bayesian modeling and genetic algorithms. Deployed both services and their dependencies on the company’s AWS Kubernetes cluster, ensuring scalable, reliable, and well-behaved production systems.
Trained and deployed efficient GAN-based video restoration and super-resolution models. Every model shipped outperformed its predecessor in both speed and quality, delivering practical improvements in real-time video processing pipelines.
Developed computer vision solutions for inventory management in retail environments. Studied state-of-the-art depth estimation, neural radiance fields, and oriented bounding box detection; introduced a novel evaluation metric that significantly reduced the memory footprint of the network.
Education
Hire Jeronim Matijević today
Post a job and invite Jeronim, or contact them directly. No hidden commission.






