Building practical AI systems with a focus on machine learning, computer vision, explainability, and reproducible engineering.
- Artificial Intelligence Engineering graduate
- Interested in AI/ML, Computer Vision, Data Science, and Software Engineering
- Experience with Python, PyTorch, scikit-learn, OpenCV, SQL, and data pipelines
- Focused on building projects that are not only functional, but also testable, reproducible, and well-documented
- Currently expanding my portfolio with production-oriented AI projects
Deep learning pipeline for breast cancer metastasis classification in lymph node histopathology images.
- ResNet-18 transfer learning
- Grad-CAM explainability
- ROC-AUC: 96.63%
- Recall: 97.67%
- 16 automated tests
- GitHub Actions CI
End-to-end machine learning system for detecting fraudulent transactions using historical customer and terminal behavior.
- Leakage-aware behavioral feature engineering
- Chronological train / validation / test evaluation
- HistGradientBoosting classifier
- PR-AUC: 92.87%
- F1 Score: 90.77%
- FastAPI inference service
- Dockerized deployment
- 45 automated tests and GitHub Actions CI
End-to-end machine learning system for predicting industrial machine failures from operational and sensor measurements.
- Random Forest with validation-optimized decision threshold
- Feature engineering and leakage-safe ML pipeline
- F1 Score: 92.63%
- ROC-AUC: 98.69%
- FastAPI inference service
- 24 automated tests and GitHub Actions CI
Real-time computer vision system for personalized ergonomic posture monitoring using a standard webcam.
- MediaPipe Pose and Face Landmarkers
- Personalized posture calibration
- Real-time multi-feature posture assessment
- Actionable live feedback and session analytics
- Persistent JSONL session logging
- 170 automated tests and GitHub Actions CI
Data & ML: NumPy, pandas, scikit-learn, PyTorch, OpenCV
Databases & Tools: SQL, Git, GitHub Actions, Docker
Focus Areas: Machine Learning, Computer Vision, Explainable AI, Data Processing
- GitHub: @ardaatikk
- LinkedIn: Arda Atik
- Email: ardacematik@gmail.com
