Résumé

• Built Flask backend control systems for Mantis, a field-deployed crop-scanning robot using RGB and thermal imaging.
• Implemented state-driven camera workflows, image capture, streaming, structured logging, error handling, and remote recovery paths.
• Integrated AWS services for media and production operations; optimized backend workflows to reduce latency by approximately 30%.

• Designed a reproducible Demand Responsive Transit platform connecting scenario configuration, simulation, policy selection, passenger response, metrics, and MLflow.
• Completed 900 simulation runs across 45 scenario variants; verified results showed learning policies improving wait-time and service-rate direction over a greedy baseline.
• Built event-driven lifecycles, dispatch-policy interfaces, experiment orchestration, and analysis pipelines.

Built a public synthetic-data ingestion reference with OIDC/JWT authorization, asynchronous validation, idempotent persistence, retries, dead-letter handling, observability, tests, and CI.

Built and operate a live Belize property discovery and operations platform with real inventory, buyer qualification, CRM workflows, parcel geometry, provenance boundaries, and consent-aware measurement.

Digitized scheduling, employee check-in, shifts, and payroll support for a Belize tour operator, reducing reported accounting errors by approximately 40%.

Normalized and analyzed 10,078 platform screen door failures across 279 stations and 18 lines to compare maintenance strategies.