NIKOLAOS SINTORIS

Fine-tuned RF-DETR on 1M+ production samples (0.979 mAP@50, 0.963 macro F1) and distilled it for faster inference. Built temporal filters that cut wind-turbine blade false alarms by 81% without reducing bird detection. Deployed automated fiber-optic inspection using YOLOv11 and Swin-T (0.91 mAP@50:95, 0.82 F1). Developed hard-negative mining, active-learning, and VLM-assisted labeling pipelines (Qwen3-VL/Ollama), with MLflow experiment tracking.

Fine-tuned YOLO for segmenting plastic types in automated waste systems. Fine-tuned ResNet50 to classify excavator states in bauxite transport. Built AI system for MCI response: triaged victims, estimated respiratory rate from ECG, and forecasted vital signs. Used YOLO for human detection/classification at low altitudes. Applied CSRNet for high-altitude crowd estimation.

- Exploratory Data Analysis (EDA) - Experience with Machine Learning Algorithms - Machine learning model deployment using Python Flask

Basic configuration on Mikrotik routers Setup an antenna for wireless network Basic skills on HTML5, CSS, PHP7, MySQL LAMP(Linux-Apache-MySQL-PHP) setup and usage of vi