Ali Murat Keceli

- Built an advice platform for open world Zelda players based on real feedback from users through reddit.

- AI assisted FastAPI + RAG pipeline with game-aware query normalization, semantic similarity thresholds, automatic live scrape→embed→requery fallback, and cached results for instant summaries.

- Multithreaded scraping reddit via praw API and duckduckgo search.

- Built a stock price prediction pipeline using PyTorch, starting with Linear Regression and advancing to LSTM networks for time-series modeling.

- Implemented a binary return classifier to forecast directional movement (up/down) and explored the limits of model performance on noisy, near-random stock data through hyperparameter tuning and evaluation.

- Achieved significant accuracy improvements through hyperparameter tuning and model optimization techniques.

- Developed a full-stack food delivery web apps using Java Spring Boot and PostgreSQL with full CRUD operations for orders and inventory, deployed with role-based authorization separating customer and admin privileges.

- Crafted RESTful controllers, service layers, and data models to ensure clean architecture, maintainability, and efficient backend logic.

- Received a 6/7 in IB Computer Science HL.

- Built a Python FastAPI and SQLAlchemy-based to-do list website with user authentication and deployed it on an Amazon EC2 instance.

- Architected secure user registration, login, and session handling to protect user data.

- Designed RESTful endpoints with proper validation and error handling, enabling clean CRUD operations for tasks.