Vasya Rusyn

I am an AI/ML engineer with hands-on experience building real-time computer vision systems, model training pipelines, and production backends. I specialize in detection and recognition solutions using PyTorch, ONNX, and modern YOLO/ResNet-based approaches.

At SAYLESS I developed real-time user and parcel tracking systems integrating detection, recognition, segmentation, measurement and OCR, and optimized inference with TorchScript and Redis-based asynchronous architectures. I build full project flows including data labeling, training, deployment and inter-service communication.

My bachelor's work produced a state-of-the-art kanji recognition model, resulted in a published MDPI article, a deployed Gradio/Hugging Face app, and an open-source Python package. I also implemented RAG-based LLM applications and full-stack web projects combining Flask/React and PostgreSQL.

I value continuous self-improvement, practical results, and building reliable systems that move research into production. I seek roles where I can contribute to CV/NLP/ML engineering, optimize inference, and grow in GPU programming, RL and scalable ML architectures.