Computer Science @ University of Vienna Β· AI / Quantum Computing / Software Engineering / Scientific Computing
Building toward a long-term goal: applying my skills in ML/AI, Quantum Computing, Scientific Computing, and Software Engineering to humanity's hardest problems - one project, one job, one paper at a time. Working through both academic research and Engineeringβdriven ventures, with the patience to build real depth before reaching for scale.
- π B.Sc. Computer Science student at the University of Vienna (ranked 1st in Austria, Top 100 globally - EduRank 2026)
- π§ Former GenAI Software Development Intern at IBM π built RAG evaluation frameworks & code analysis tools
- βοΈ Former Software Development Intern at Bosch π ML, data cleaning, tool migration
- βοΈ Quantum Computing Research Intern at QWorld π quantum image encoding (QPIE, FRQI), Quantum Harris Corner Detection
- π 3rd Place & Google DeepMind Special Recognition π AI Agent Olympics Hackathon, Milan AI Week 2026
- πΉπ· TEKNOFEST Top 20 β Turkish NLP Scenario Category
| Project | Description |
|---|---|
| OmniSynth π | Full-stack multi-agent AI research platform turning collections of papers into an interactive knowledge workspace (concept graphs, per-paper wikis, gap analysis, hypotheses). 3rd Place & Google DeepMind Special Recognition at AI Agent Olympics Hackathon, Milan AI Week 2026. |
| RAG Testing Tool [WITHIN IBM] | A synthetic-data-powered testing & evaluation framework for RAG systems, built with Langfuse, WatsonX, Ragas, Streamlit, and FastAPI. Built during IBM internship. |
| Quantum Harris Corner Detection | Quantum realization of Harris/Sobel-based corner & edge detection using FRQI and QPIE image encoding in Qiskit, with classical post-processing. |
| Agentic Turkish LLM (TEKNOFEST) | Multi-agent sentiment analysis system combining a fine-tuned Mistral-7B classifier with a LLaMA 3.1 8B orchestrator, trained on TΓBΔ°TAK HPC resources. |
| Hello HPC | Beginner-friendly tutorial and toolkit bridging classical HPC and quantum workflows, using the Qiskit 2.3 C API, Slurm, and OpenMP. |
| Repro | Reproducibility Co-Pilot: an AI-assisted tool that turns a paper + its repo into a step-by-step, human-approved reproduction plan, executed autonomously in an isolated Docker sandbox with LLM-driven replanning on failed steps. |
| BenchPilot | Converts a one-sentence research hypothesis into a full, literature-grounded experiment plan β automated novelty checks, source-grounded methodology generation, and a scientist review loop that remembers corrections across sessions. |
| NeurosymbolicSLM | Solo research project (NeSy 2026 submission) testing whether a 7B math-specialized model with tool-integrated reasoning can match GPT-4o on GSM8K β a resumable, statistically rigorous experimental pipeline with a transparent write-up of where the paper's claims didn't survive peer review. |
| iac-26-transformers | Research probe testing whether transformer models can learn meaningful latent representations of periodic orbit trajectories in the Earth-Moon system (CR3BP), benchmarked against MLP/CNN baselines. |
- Qiskit Global Summer School 2025 β Quantum Excellence
- IBM watsonx.ai Technical Essentials Β· 2025 IBMer watsonx Challenge
- MIT iQuHack 2025
- DeepLearning.AI TensorFlow Developer Specialization (4 courses)