Master's Student in Applied Physics with a Focus on Quantum Computing Hardware
Contact Me View CV
I am a passionate and dedicated physicist with a strong interest in quantum hardware, quantum computing and machine learning. I have experience in research, teaching, and leading projects. I am eager to contribute to research and advancements in these fields.
MSc Applied Physics, Delft, Netherlands (2023 - Present)
Expected Graduation: June 2025
BSc Physics, Istanbul, Turkey (2018 - 2023)
GPA: 3.16
Gaziantep, Turkey (2014 - 2018)
GPA: 99.87/100 | Valedictorian
Jun 2020 - Mar 2021
Jan 2020 – June 2020
June 2019 – Current
Feb 2020 – Jan 2022
Nov 2022 – Jan 2024
Gutenberg Corpus, StyleSet, and StyleBART focuses on creating a large-scale dataset for author style transfer and classification. Utilizing GPT-3.5 for generating a pseudo-parallel dataset and models like BERT and BART, the project involved scraping and preprocessing 119M sentences from 10 authors in Project Gutenberg. I fine-tuned BERT for author classification, achieving 60% accuracy, and BART for style transfer, though it faced training challenges. This project highlights innovative approaches in model fine-tuning and dataset creation for high-fidelity style transfer and authorship classification, demonstrating the potential for further advancements in natural language processing.
Learn More
QKART is an educational mobile game that introduces users to quantum programming through interactive gameplay. Developed with Unity Engine and Blender for 3D modeling, the game uses X, Z, H, and CNOT quantum gates to manipulate two qubits, simulating quantum superposition and measurement by altering the probability of the player's car being in different lanes. C# scripting simplifies complex quantum concepts, making them accessible to high school students and above. The game aims to provide an intuitive way to understand quantum computing, encourage hands-on learning, and promote replayability to reinforce knowledge. Recognized for its educational value, QuTie won first prize in the "Education and Outreach Materials" category at a Quantum Technologies Hackathon. Project received recognition from IBM.
Learn More
Using Decision Transformer in Drone Environment employs advanced reinforcement learning (RL) techniques and transformer models to enhance drone navigation. Utilizing the Decision Transformer, Soft Actor-Critic (SAC) algorithm, and PyBullet simulation, the project aims to reproduce RL baseline results, train SAC agents for precise drone movement, and generate robust datasets. Outcomes include a trained Decision Transformer agent that significantly improves drone control in complex environments and an effective dataset generation process. This project highlights the potential of transformer models in advanced RL applications.
"Dancing2Music In Style" explores the latent space between music and dance to generate realistic dance choreographies from audio inputs. Using technologies like Generative Adversarial Networks (GANs), Variational Autoencoder (VAE), PCA, and T-SNE, the project collected and preprocessed dance videos, modified the Dancing2Music model for latent space manipulation, and analyzed these spaces. The outcomes demonstrated successful cross-modal latent space manipulation, producing dance sequences with characteristics of target styles like Ballet and Zumba, highlighting the potential for further research in cross-modal generation.
Learn More
Developed and implemented Quantum Finite Automata (QFA) algorithms using Qiskit to solve the MODp problem on IBM's noisy quantum devices, focusing primarily on MOD11. This project aimed to create highly efficient quantum circuits, resulting in a compact 3-qubit solution with reduced gate count. By leveraging the Moore-Crutchfield QFA model, we explored the potential of quantum computing for significant space savings over classical methods. Through extensive testing on both simulators and real quantum hardware, we gained valuable insights into optimizing performance despite the challenges posed by quantum noise. The project also experimented with larger instances like MOD31,
Learn More
This project investigates metal deposition and liftoff processes using a dithering mask to enhance nanostructure to test binary 3D fabrication. Techniques employed include sputter deposition of Au and Ti under standardized conditions, and varying resist thicknesses and profiles, and testing methods such as argon milling. The study utilized advanced characterization methods such as Scanning Electron Microscopy (SEM), Optical Profilometry, and White Light Spectrometry to inspect and quantify deposition outcomes. By controlling variables like metal type, processing, and mask patterns, we systematically analyzed parameters affecting resolution, profile quality, success rate, and processing time. This comprehensive approach demonstrates expertise in nanofabrication processes and advanced material characterization techniques.
SoonQTurkey
Issued: Oct 2020