Projects

In this brain-computer interface project I developed, I detected users' movement intentions (like 'forward' or 'left') using brain signals from dry-electrode EEG systems. While processing and filtering signals with MNE and Matlab, I developed deep learning models based on TensorFlow and PyTorch to generate 2D commands from raw EEG data. The system also has the ability to detect and correct its erroneous predictions thanks to brain-based feedback (ErrP). The project was completed with a user-friendly interface that allows motor-impaired individuals to control a simulated wheelchair or robotic arm.

A modern financial tracking application. Users can add, edit, and track the performance of their portfolios or stocks via graphs. The project is built on the MERN (MongoDB, Express, React, Node.js) stack, and the Vite framework was used for the frontend.

A desktop artificial intelligence application I developed using Electron and React to help companies and public institutions easily integrate AI into their workflows to increase efficiency and prevent data from being processed locally and leaving the company. It allows users to use online or local AI models without needing technical knowledge.

To address the lack of Turkish chatbots, I trained the Google Gemma model on Google Colab over a 40-hour process in fragments using the Unsloth library and LoRA technique; I created a dataset of 400,000 Turkish Wikipedia articles and 50,000 law, education, and agriculture-focused question-answer pairs, developed 4 different models, and shared them on the HuggingFace platform for public access.

I took part in a 5-person team on a European Union-supported project. I directed the quality assurance (QA) processes, prepared SRS and SDD documents, and created and executed comprehensive test plans and test cases.

I created a real-time 2D simulation environment with Unity and developed a game in C#. In this environment, I trained multiple agents using Python-based RL libraries and comparatively analyzed their learning processes and adaptation abilities by testing each agent under different, dynamic environmental conditions. In the project, I demonstrated the potential of Reinforcement Learning algorithms in critical application areas like the defense industry by measuring the agents' performance in automation scenarios.

I designed an artificial intelligence model using Python and Yfinance that automatically collects and processes financial data, analyzes market movements, and provides strategic insights to the user. Thanks to the model I developed, even users with limited financial knowledge can make informed investment decisions by receiving personalized recommendations on products like stocks and mutual funds.