Gjorgji Strezoski

Advisor: Marcel Worring (Informatics Institute)

GPA 9.9 / 10.0

GPA 9.1 / 10.0

Leading a team of 3 engineers on the Data & ML direction, building practical systems to deliver insights in real time. Work includes multi-model recommenders using pruned transformers and statistical models, data platform engineering for ML-ready pipelines (Databricks, PyTorch, ggml, vLLM), and internal data tool development and deployment.

Laid the foundations of New Black's data platform, designing core architecture for reliable, scalable data use. Implemented the medallion architecture across the organization, delivered end-to-end data pipelines from ingestion to live dashboards, and established the backbone for analytics, reporting, and machine learning capabilities.

Building high-reliability systems and cost-efficient software solutions for e-commerce and logistics.

Design of user interactions with an LLM through visual and conversational alignment, building a mental health support agent powered by custom RAG flows and a medical knowledge database.

Designed and developed an international shipping platform software that converted regular stores into fulfilment warehouses for greener more efficient shipping. Led an interdisciplinary team of engineers and designers to scale across 15 countries in less than 10 months. Exited in March 2024.

Founded a software development company with 5 engineers specializing in Django and surveillance image stitching. Design and develop bespoke applications for web and mobile platforms.

Collaboration on scientific research in the computer vision, machine learning and multimedia domains.

Built a realtime instance segmentation engine for a large dataset of rigid objects under diverse controlled lighting conditions for training a model that powers such characters a AR game.

Developed attribute-driven deep models for representing semantics and contextual understanding behind Frank van der Salm's work for the N_O_W_H_E_R_E exhibition.

Research partner for the VISTORY project at the premier art museum of the Netherlands, applying computer vision to the museum's collection. Hosted and participated in interdisciplinary colloquiums with professionals from diverse domains. Work resulted in scientific publications and graduated master students in AI and Computer Vision.

Rails backend development and machine learning on proprietary data.

Machine learning, deep learning, image processing and data science research as part of the MAESTRA Project.

G. Strezoski, N. van Noord, M. Worring, "MATTE: Multi-task multi-scale attention", Computer Vision and Image Understanding (CVIU), 2023

G. Strezoski, N. Van Noord, M. Worring, "Learning Task Relatedness in Multi-Task Learning for Images in Context", International Conference on Multimedia Retrieval (ICMR), 2019

G. Strezoski, N. Van Noord, M. Worring, "Many Task Learning with Task Routing", IEEE International Conference on Computer Vision (ICCV), 2019 (Oral)

G. Strezoski, M. Worring, "OmniArt: A Large-scale Artistic Benchmark", ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 2018

G. Strezoski, I. Groenen, J. Besenbruch, M. Worring, "ArtSight: An Artistic Data Exploration Engine", Demo Track at ACM Multimedia Conference (ACM MM), 2018

G. Strezoski, I. Groenen, J. Besenbruch, M. Worring, "Plug-and-Play Interactive Deep Network Visualization", Visual Analytics for Deep Learning Workshop at IEEE VAST, 2017

Generative Adversarial Networks for Decorative Initial Image Generation

Jointly supervised with Nanne van Noord

Multi-task Initial Classification

Jointly supervised with Nanne van Noord

Art Inspired Fashion with Generative Adversarial Networks

Jointly supervised with Nanne van Noord

Facial Sentiment Analysis in Artistic Painting