Samuel User

Supported RLH research to improve large language models by refining reasoning quality and problem-solving coherence. Curated supervised fine-tuning datasets with Pandas, evaluated frontier models in simulated multi-agent environments, and designed adversarial scenarios to stress-test agent performance.

Developed and maintained full-stack web applications using React on the frontend and NestJS with MySQL on the backend. Automated parts of a data-collection workflow with a Python script, supported production runs for multiple datasets, and managed remote deployments and client integrations.

Built a full-stack application to track cow estrus cycles and provide timely alerts for farmers, inseminators, and veterinarians in Debrezeit, Ethiopia. Implemented automated notifications using the AfroMessage API and supported an MVP pilot with training of users.

Developed cross-platform UI using Flutter for RateEat restaurant item ratings. Integrated the mobile frontend with a Node.js backend and coordinated sprint activities through collaboration and problem-solving.

Built deep learning models for coffee leaf segmentation and disease detection using U-Net and OpenCV. Improved robustness through data preprocessing, augmentation, and dataset diversification, and prepared performance reports and presentations.

Activities and societies: Relevant coursework included Data Structures & Algorithms, Web Development, Database Design, Computer Architecture, Operating Systems, and Programming Languages.

BSc in Software Engineering, covering core topics in software development and computer science foundations.