Kadir Çağrı Kuzaytepe

Teknofest International UAV Competition – Free Task Category (2024-2025)

As the captain of the 16-member Antlers UAV team, I led efforts to count vehicles and pedestrians at intersections and conduct analyses for traffic signals. In the project, UAV control was implemented using the Pixhawk 4 flight controller, while I developed control algorithms using Python and Thinker. In the field of image processing, I created a specialized dataset and trained a model for vehicle and pedestrian counting using the YOLOV11 model. Additionally, I developed a Flutter-based mobile application and integrated FastAPI to enable mobile UAV control. This project not only enhanced my leadership and project management skills but also provided me with in-depth experience in image processing and flight algorithms.

Pack & Eat – Food Delivery Web Application

• Developed end-to-end multi-role platform (customer, restaurant, courier, admin) using ASP.NET Core 8 with MVC architecture and SQL Server + EF Core back-end.
• Built responsive React.js frontend with Tailwind CSS; implemented JWT-based security, live courier tracking via Google Maps API, and Supabase for image storage.
• Collaborated on GitLab with CI/CD pipelines, ensuring maintainable version control; API reliability validated via Swagger and Postman testing.
• Gained full software lifecycle experience: collaborative system design, coding, testing, and deployment—mirroring industry workflows.

Machine Learning Projects – Sentiment Analysis & Image Classification

• I developed an LSTM-based sentiment analysis model to classify IMDb reviews and strengthened its semantic understanding using FastText embeddings. I compared a customized LSTM model with PyTorch LSTM, evaluating performance through metrics such as accuracy and F1-score.
• For the Animal-10 dataset, I created an image classification system by designing a CNN from scratch and employing transfer learning methods with ResNet50 and MobileNetV2. Through data augmentation and preprocessing techniques, I enhanced the models’ overall performance and achieved the highest accuracy level with transfer learning.

Self-RAG Agent for Secure Document Analysis

• Architected a fully local, offline Retrieval-Augmented Generation (RAG) system to provide secure, private intelligence from sensitive documents without external API calls.
• Engineered an advanced agentic workflow using LangGraph, enabling the system to perform self-correction, query rewriting, and final self-critique to generate 100% accurate, hallucination-free answers.
• Built a custom data pipeline featuring semantic chunking and indexed document vectors in a locally deployed Vespa instance using Sentence Transformers for fast, on-premise semantic search.
• Integrated a quantized Llama model, cloud-free text generation, optimized for thread-safe concurrent queries.
• Developed a custom Streamlit interface to serve as a complete control panel for the system, from document indexing to real-time performance monitoring.

ARK4UTH Authenticator Application

• Engineered a secure mobile authenticator application with a responsive frontend built in Flutter.
• The application's backend is powered by a robust .NET and MSSQL infrastructure, designed with a multi-layered architecture for scalability and maintainability.
• Features end-to-end data security, utilizing strong AES encryption algorithms to protect user credentials and data in transit.
• High performance and reliability are ensured through strategic use of Redis Cache for rapid data access and comprehensive logging for traceability.
• Includes secure API integrations for seamless data flow and a push notification system to deliver real-time alerts and enhance user engagement.

Hacettepe University Kristal Geyik Awards Website Update

As part of the Hacettepe University Kristal Geyik Awards 2025, I served as a software developer responsible for updating the website. In the project I implemented using React, I optimized the user experience by applying modern and responsive design principles.

Python Instructor

During the 2024–2025 academic year, I led a four-week Python training session as part of the ProgrammixV event, offering a comprehensive curriculum that spanned from basic programming concepts to advanced Python features. Throughout the course, I provided lecture notes, assignments, quizzes, and sample projects, allowing participants to practice and reinforce their theoretical knowledge. In addition to core programming fundamentals, I covered topics such as data structures, functions, error handling, and file operations, guiding participants in advancing their programming skills