Generative Image AI
Neural Style Studio: Custom SD 1.5 LoRA Engine
Applies the selected art movement to uploaded images using lightweight LoRA adapters trained for Cubism, Post-Impressionism, and Ukiyo-e.


AI and Data Portfolio
3rd-year Artificial Intelligence and Data Engineering student at Fırat University. Focused on architecting production-ready AI systems, automated NLP pipelines, and practical tools that make complex models truly usable.
Location
Fırat University, Turkey
Tech Stack
Python, Scikit-learn, PyTorch, TensorFlow, FastAPI, NLP, Image Processing
Currently Learning
RAG, NLP, image processing, and practical AI products
Profile
Specializing in transforming cutting-edge deep learning models into practical, automated software systems. Focused on building usable, production-ready NLP and RAG pipelines that solve real-world problems.
ML/DL Foundation
Solid grounding in core machine learning and deep learning architectures. Emphasizing rigorous experimentation, robust model evaluation, and high-performance inference pipelines.
Current Areas
Engineering end-to-end AI workflows that transform isolated experiments into fully automated, usable software systems. Focus areas include advanced NLP, computer vision, and scalable RAG architectures.
Engineering Philosophy
Static notebooks don't ship. I build end-to-end AI systems—from training and fine-tuning to deploying clean API endpoints and functional UIs that deliver real user value.
Work
Generative Image AI
Applies the selected art movement to uploaded images using lightweight LoRA adapters trained for Cubism, Post-Impressionism, and Ukiyo-e.
Probabilistic Deduction Engine
A noise-tolerant Bayesian deduction engine that selects atomic questions through expected information gain across a curated 115-animal knowledge base.
Natural Language Processing & RAG Systems
Production-ready system featuring two-stage intent classification and ChromaDB vector routing for high-accuracy Turkish customer NLU.

