I’m Emre, a computer engineer working with generative AI, large language models and AI agent architectures. I focus on turning models into applications that address real problems.
I developed embedding-based retrieval and explainable analysis for SafeDose and completed a RAG-based customer support assistant with a team at Microsoft Summer School. My work covered data preparation, finding relevant knowledge and supplying context to language models.
I built my machine learning and deep learning foundation with Python, PyTorch and TensorFlow. I worked on backend systems connecting models to applications through Flask, FastAPI and PostgreSQL. Today I’m extending this experience into generative AI, RAG and AI agent architectures.
During the program beginning in June 2026, I built a customizable customer support assistant with a team. The system used company documents, historical support tickets, FAQs and user manuals as knowledge sources. I developed RAG-based retrieval with Python and Qdrant for document vectorization, semantic search and contextual input to the LLM. The project is completed.
January 2026 — present. As a founder and AI / data engineer, I contribute to SafeDose. I developed embedding-based retrieval and LLM-powered explainable analysis for drug–drug and drug–food interactions in the Swift-based mobile app. The project is available on the App Store, earned second place at the AI’mpact Full-Stack AI Agent Hackathon and was accepted into ITU Seed pre-incubation.
December 2025 — February 2026. I managed content, page updates and digital editing across the university website. I helped maintain more than 100 pages and published over 50 content updates, supporting accurate information and consistent website organization.
August — September 2025. I prepared training data and labeled images with CVAT / Roboflow for YOLOv8 models, contributing to model optimization. I built a real-time face recognition attendance system with Flask and PostgreSQL, managed more than 1,000 face embeddings and created an employee tracking dashboard.
July — August 2025. I developed low-latency, real-time object detection systems using YOLOv8 and YOLOv5-OBB for industrial automation, achieving over 90% mAP. I worked with ByteTrack and OC-SORT for video tracking and compared models through precision, recall and mAP to evaluate suitable solutions.
September 2024 — February 2025. As part of the AI/ML core team, I helped plan, organize and run six AI-focused workshops attended by more than 300 students. I supported technical knowledge sharing and engagement across the developer community.
I developed embedding-based retrieval and LLM-powered explainable analysis for a mobile application analyzing drug–drug and drug–food interactions.
I built a customer support assistant with a team, grounded in company documents, support tickets and user manuals. I developed the RAG retrieval system using Python and Qdrant to vectorize documents, perform semantic search and supply relevant context to the LLM.
Learning & certifications — DeepLearning.AI · Miuul · BUSIBER Boğaziçi
My experience, projects and education — in one document.