Hi, I'm Enes Demir
Computer Engineering Student at Kocaeli University. Developing myself in Machine Learning and Data Science, aiming for a career in these fields.
About Me
Machine Learning Engineer with a focus on building scalable AI solutions.
GPA: 3.71 / 4.00
My Journey
My journey began in my first year of university, where I laid a strong foundation in Computer Engineering by mastering algorithms and programming in C. This period was dedicated to understanding the core logic of computation and problem-solving.
At the beginning of my second year, I shifted my focus towards Data Science and Machine Learning. I was fascinated by the ability to extract insights from data, which naturally led me to dive deeper into the field.
I then focused heavily on Data Science and Machine Learning, working with both image and tabular data. I specialize in handling data scarcity and conducting research on SOTA models, with a particular emphasis on classification tasks, Domain Generalization, and Self-Supervised Learning.
As AI agents gained popularity, I expanded into Agentic AI — building multi-agent orchestration systems, RAG pipelines, and exploring how LLMs can autonomously reason, plan, and execute complex tasks.
In the summer of my third year, I completed a one-month AI internship at XON Technology in Gebze Bilişim Vadisi. I built a leakage-safe scalp baldness segmentation pipeline, estimated graft counts with conformal prediction, and shipped a physician workflow with quality checks, post-op preview, and automated PDF reports.
Skills & Tools
AI & ML
- PyTorch
- TensorFlow
- Scikit-learn
- Pandas
- NumPy
Backend & Data
- FastAPI
- Flask
- Spring Boot
- SQL
- Docker
Languages & Tools
- Python
- C
- Java
- TypeScript
- JavaScript
Projects
AI & ML
GitHub
Web & Mobile
Experience
Built a patient-level, leakage-safe scalp baldness segmentation pipeline on 251 patients (191/30/30 train/val/test), comparing YOLOv8 and SegFormer and reporting Dice 0.62 and area MAE 0.034 on the held-out test set. Estimated graft count from mask geometry with ridge regression and conformal prediction intervals, validating 80% target coverage on unseen patients. Shipped a physician workflow with image upload, quality checks, post-op preview, and automated PDF reports showing graft estimate and interval.
Developing a domain generalization framework utilizing Self-Supervised Learning (SSL) to extract robust feature representations from heterogeneous Chest X-Ray datasets. Conducted extensive comparative analysis and benchmarking of SOTA architectures (Vision Transformers, ConvNeXt, Swin Transformer). Engineered a modular, professional-grade training pipeline in PyTorch featuring automatic checkpointing and experiment tracking. Implementing feature fusion architectures to combine tabular clinical data with visual embeddings.
Kocaeli University
