Engineering projects I have built or am actively building.
Central service for fleet metadata, robot configuration, and software release management. Built from the ground up: domain model, REST API, robot agent, CI/CD, database migration, and cloud deployment.
The OS upgrade across the entire production robot fleet. Built migration scripts with retry logic, backup and restore procedures, and Ansible playbooks for safe fleet-wide rollout.
Migrated CI/CD and Debian package distribution from on-premise to cloud across 30+ repositories. Built centralized GitLab CI components for SSH setup and Debian releases, meaning common workflows change in one place instead of thirty.
Master's level research applying deep learning and sensor fusion to complex engineering problems.
Master's thesis
Designed and implemented a 2DCNN for classifying and localizing damage in a two-stage planetary gearbox using vibration signals. The spectrogram-based deep learning approach outperformed random forest and SVM under all operating conditions.
Master's project — SMART2 Rail
Built a central decision support system that processes obstacle detections from multiple train-mounted sensors and a UAV. The system estimates GPS coordinates using a machine learning distance model and fuses duplicate detections via weighted averaging, completing all steps in under 8 milliseconds.