Welcome to my GitHub! I'm a graduate student in Artificial Intelligence with a background in Computer Science & Engineering. My primary interests are in AI, robotics, and VR. I work on projects that combine technical depth with practical applications, aiming to make technology more interesting, effective and inclusive.

• Large Language Models (LLMs): Fine-tuning, prompt engineering, and real-world applications.
• Robotics: Integrating AI with robotic systems for problem-solving in dynamic environments.
• VR Accessibility: Developing tools to make virtual reality environments accessible to users with low vision.
• Deep Learning: Applying neural networks in reinforcement learning, generative models, and computer vision.
• High-Performance Computing: Optimizing AI workflows with parallel and distributed systems.

• MagniVR:

  • Designed a VR accessibility SDK for low-vision users, making it easy to integrate with existing VR applications.
  • Published at HCII 2024 and received the Universal Design in Information Technology Award at the INTERNATIONAL BARRIER-FREE INFORMATION TECHNOLOGY CONGRESS 2024.
  • Conducted a two-phase user study with 18 participants, demonstrating its usability and inclusivity.

• Chaining Gans Project:

  • Explored semantic face editing by chaining GANs, using SVM-trained boundaries in latent spaces for precise attribute control.
  • Achieved improved fidelity and style preservation through innovative latent space manipulation.

• Auto Space Shooter with ML Agents:

  • Used deep reinforcement learning and curriculum learning to train AI agents for progressively challenging game tasks.
  • Compared performance of self-taught agents with those trained using human imitation.

• LLMs for HPC: Developed a pipeline to detect performance bottlenecks in HPC code using LLMs and prompt engineering.
• Electrospinning Automation: Automated a 4-axis robotic system for nanofiber production with electronic and kinematic control.
• IoT Biometric Security System: Built a face and speech recognition-based authentication system using edge devices.
• Ray Tracer: Created a photorealistic CPU ray tracer supporting advanced lighting and materials.
• Can Wait - The Game: Designed and developed a procedural roguelike networked co-op game with Unity.

Programming Languages: Python, C++, CUDA, C#, Go, Rust, JavaScript, Solidity, Verilog, Scheme, MIPS Assembly
Frameworks & Tools: TensorFlow, PyTorch, OpenCV, OpenGL, Unity, Unreal Engine, Docker, AWS
Specializations: Generative AI, High-Performance Computing, VR Development, Robotics, Blockchain

• Website: cem-kaya.github.io
• GitHub: Cem-Kaya
• LinkedIn: Cem Kaya
• Google Scholar: My Publications