Computer Engineering student at Erciyes University, Türkiye. I build machine-learning and optimization tools that can be trusted: measured uncertainty, fair comparisons, reproducible results. My work spans swarm-intelligence optimization (the Artificial Bee Colony algorithm), predictive maintenance for aerospace, and ML for particle physics as an undergraduate researcher in the Erciyes Neutrino Research Group, a member institution of the ANNIE (Fermilab) and EOS (UC Berkeley) experiments.

aero-rul How many flights does this engine have left, and how far can that answer be trusted? Remaining-useful-life prediction for turbofan engines on NASA's N-CMAPSS data, with conformal prediction intervals whose coverage is measured on engines the model has never seen. pip install aerorul 90 % interval covers 90.4 % of test flights; plain quantile regression delivers 82.2 % 10.1 flights RMSE, engine-level splits, no label leakage, five seeds

abcopt Canonical Artificial Bee Colony optimization for Python, with ABCSearchCV, a drop-in alternative to scikit-learn's GridSearchCV. pip install abcopt Verified step for step against the Erciyes University reference implementation Continuous, integer and categorical search spaces, cv_results_, caching 54 tests, CI on Python 3.9–3.12, benchmark against Optuna TPE

• Erciyes Neutrino Research Group · Undergraduate Researcher, TÜBİTAK 2247-C scholar Cherenkov / scintillation photon separation in water-based liquid scintillator detectors: ROOT-to-ML data pipelines and a benchmark of 20 classifiers, where a tuned ensemble reached 95.75% accuracy versus ~90% for the classical time-cut method. Administrator of the group's Linux CPU/GPU servers.
• TÜBİTAK 2209-A · Project Leader Hyperparameter optimization of photon-separation models with ABC, compared with Grid Search, Random Search and Optuna TPE under equal budgets (code & results).
• POSCO Assan TST Steel · Data Science & Software Engineering Intern Defect-driver analysis on production sensor data (XGBoost, GroupKFold, SHAP) and a YOLOv8 segmentation model for steel surface defects.
• Graduation thesis: KOVAN · in progress LLM-guided Artificial Bee Colony for discovering interpretable algorithms, with applications in optimization and particle physics.

• CS50x Türkiye · core team member
• NASA Space Apps Challenge Kayseri · mentor (November 2026)

• Languages: Python · TypeScript · Java · SQL · C/C++
• ML & optimization: PyTorch · scikit-learn · XGBoost · LightGBM · Optuna · SHAP · Ultralytics YOLOv8 · conformal prediction · metaheuristics (ABC)
• Data & tools: NumPy · Pandas · ROOT (CERN) · PostgreSQL · Git · GitHub Actions · MkDocs · Streamlit · React · Gemini API · Linux server administration

• Building KOVAN for my graduation thesis
• Preparing for a research-oriented Master's in AI/ML in Germany
• Former Erasmus+ student at Częstochowa University of Technology, Poland

Website · LinkedIn · Email · X

_{Working to give direction to science and technology, from the present to the future.}