Backend (core platform team)
•Integrated a YOLOv8-based smoke and fire detection ML model with 95.7% precision, satisfying a major customer
requirement and securing their contract renewal.
•Co-led the redesign of legacy backend services from a misfit NoSQL/TypeScript/Python on AWS Lambda stack to
a Go/PostgreSQL/Kubernetes architecture, correcting schema flaws with a relational model, improving system
flexibility, enabling cloud-agnostic deployments, and reducing REST API latency by 42%.
•Designed a hybrid migration strategy that preserved legacy REST endpoints while rerouting them via gRPC to new
Go services, enabling partial adoption during the transition and overcoming challenges in schema mismatches, user
permission mappings, and unknown edge cases.
•Designed and built a location telemetry system with TimescaleDB, introducing efficient spatial queries by proximity
and metrics. This unlocked extensibility for features like drone integration and threat localization, ultimately
contributing to securing an $8M DoD customer contract.
•Architected an event streaming system (built on NATS) to aggregate and filter organizational events by user
interests, enabling timely and resilient notifications in disconnected, intermittent, and limited (DIL) environments.
•Architected a customizable notification system with SMS (AWS SNS) and email (Go SMTP) delivery, empowering
users to control frequency and scheduling preferences.
•Designed and delivered a self-service command and control (C2) system using MQTT and client-defined schemas,
enabling hardware and software clients to register their own command structures. This removed the server as the
single source of truth and supported resilient operations in disconnected, intermittent, and limited (DIL) environments.