I am a BSc (Hons) Computer Science graduate from the University of Manchester and a Robert College of Istanbul graduate. My work spans neural model training, pretrained transformer fine-tuning, classification pipelines and experimental evaluation.
I am interested in Software Engineering, AI/ML Engineering, NLP, Research Engineering and Data Science opportunities, primarily in Türkiye.
-
Biomedical Citation Validity Analysis — For my dissertation, I prepared claim–evidence data, implemented classical ML baselines and fine-tuned pretrained PubMedBERT, comparing it with GPT-3.5 prompting. The repository includes an experimental Streamlit interface; it is a research prototype, not a validated production application.
-
Multi-Label Movie Genre Classification — Built preprocessing, training and prediction pipelines for nine genres. Trained a BiLSTM from randomly initialised parameters using Keras layers and separately fine-tuned pretrained RoBERTa; compared recovered validation results using label-aware metrics.
-
CIFAR-100 CNN Hyperparameter Study — Extended a teaching CNN foundation with a 100-class Keras pipeline, augmentation and regularisation. Trained models and compared learning rates, batch sizes and dropout, with saved learning curves and documented evaluation limitations.
-
Air Quality ML and Optimisation — Implemented NumPy batch-gradient updates, prediction and classification metrics within teaching scaffolding. Compared learning rates and scikit-learn MLP regression configurations, documenting preprocessing and model-selection limitations.
These academic repositories preserve historical implementations and results. Their READMEs distinguish reported findings, retained outputs and reproduction limits.
- Gender Bias in Word Embeddings — Analysed pretrained GloVe associations using cosine similarity, descriptive contrasts and visualisations.
- Digital Carbon Footprint — Tableau visualisation of historical emissions assumptions, with explicit methodology and limitations.
- Programming: Python, Java, C++, SQL
- ML and data: TensorFlow/Keras, PyTorch, Hugging Face Transformers, scikit-learn, NumPy, pandas; NLP, classification and model evaluation
- Software tools: Git, Streamlit, API integration