Muhammet Furkan Kupcu

• I developed a YOLO-based license plate recognition system that achieved 90% mAP on the validation dataset.
• I deployed deep learning models to real-time environments using OpenCV-DNN and Docker, increasing system efficiency and scalability.

• I learned the basics of IT support, networking, and hardware systems.

I developed an interactive AI assistant aimed at improving mental well-being. The project leverages modern LLMs and fine-tuning techniques to provide supportive and empathetic conversational experiences.

I developed a robust pose estimation pipeline for traditional folk dances using a curated dataset and fine-tuned models.

I implemented a movie recommendation system using Graph Neural Networks (GNN) in PyTorch to model user-movie interactions as a bipartite graph and predict user ratings through learned graph embeddings.

I developed a video object detection tool using the RF-DETR model to detect specific object classes with configurable parameters (threshold, FPS, etc.) and generate cropped and annotated videos.

An award-winning card game where player choices matter. As the leader of your country, you must make strategic choices to maintain a sustainable system and prepare a better future for generations to come. Funded in part by a grant from the U.S. Mission to Turkey.

A short 2D point & click hand-drawn puzzle game. The story follows Lara, a girl lost in forgotten ruins, and her brother's spirit, Aiden, helping her escape. Players switch between the world of the living and the spirit world to solve puzzles.