At Koç Healthcare – American Hospital, I designed a multi-agent software development workflow with planner, orchestrator, developer, tester, and reviewer roles. I also designed API contracts and implemented Python and Quart endpoints for an internal log monitoring system.
At Jettnet Bilişim Hizmetleri, I contributed to an AI-powered student recognition system, designing PostgreSQL structures and developing FastAPI services that connected recognition outputs with application data.
At AISenseUs, I developed a multimodal emotion recognition platform for visual, video, and audio inputs. I trained deep learning models and integrated a FastAPI backend with a React frontend, achieving up to 91% accuracy on video emotion recognition and 86% on audio classification.
For my undergraduate thesis, I developed CrisisAid-TR, an AI-assisted disaster response platform using a Turkish BERT model. The project achieved 82% accuracy and 0.86 F1 on the aid class, was supported by TÜBİTAK, and received second place in the undergraduate thesis competition.

