BDDK RAG System
A RAG system built on BDDK (Turkish banking regulator) compliance documents for banking and fintech use cases. Developed during the Microsoft Cloud Summer School internship; work is still ongoing.
Backend Developer · AI & LLM Integration
4th-year Computer Engineering student at İnönü University. I work on backend development and AI/LLM systems.
6+
Projects
15+
Technologies
10+
GitHub Repos
01 — About
I'm a 4th-year Computer Engineering student at İnönü University, working at the intersection of backend development and AI/LLM systems (RAG, prompt engineering). I gained leadership experience within the IEEE and Google DSC communities.
02 — Experience
2023
Started my Computer Engineering degree.
2023 — 2024
Took part in event organization and technical content creation.
2024 — 2025
Contributed to the planning and execution of workshops and hackathons; took part in forming and coordinating project groups.
2025 — 2026
Led technical event planning and internal team coordination; owned member development processes.
2026
Built a RAG assistant for BDDK compliance documents using Phi-3.5 Mini, Foundry Local, and PostgreSQL; implemented the full pipeline end-to-end: document chunking, embedding, retrieval, and response generation.
03 — Projects
Keep scrolling
A RAG system built on BDDK (Turkish banking regulator) compliance documents for banking and fintech use cases. Developed during the Microsoft Cloud Summer School internship; work is still ongoing.
A Random Forest classifier built on a pan-cancer RNA-Seq dataset with 801 samples, 20,531 genes, and 5 cancer types. SHAP-based explainability added; the model achieved ~99.4% accuracy.
An iOS app bringing together prayer times, supplications, Quran, and a dhikr counter. Published on the App Store; free with an ad-supported model and a premium system ready to be enabled.
A personalized Turkey travel planner featuring custom route planning and interactive map integration. The web version is complete; a mobile version is in development.
An NLP project detecting adverse drug reactions from Turkish social media data. A BERTurk model was fine-tuned for classification.
The site you're currently on. A personal portfolio built with Next.js — bilingual (TR/EN), featuring a Three.js particle background and GSAP animations. Deployed on Vercel with SEO, security headers, and accessibility in mind.
04 — Skills
05 — Contact
Feel free to email me about internships and entry-level opportunities — or just to say hi. I'll get back to you as soon as I can.