O3 · March 2026–present
Shipyard Operations
Software
Deployed software
HII shipbuilding use case
An operational ontology that connects the moving parts of shipbuilding.
I built and deployed shipyard-operations software for O3 using Rust and Python, designed around HII shipbuilding workflows. Its ontology connects parts, equipment, workers, and task dependencies. Agentic coordination and constrained scheduling turn that shared state into planning insights and coordinated responses to changing production conditions.
- Saronic qualification
- Software engineering in Rust and Python
- Saronic duty
- Adapting software to operator requirements from direct feedback
TechnologyRust · Python · FastAPI · PostgreSQL · React · TypeScript · OPC-UA · OR-Tools · Docker
- My responsibility
- Ontology, service integration, agentic planning workflow, operator interface, and local deployment.
- Operational outcome
- Surface bottlenecks and coordinate interdependent work to help prevent delays in fast-moving shipbuilding environments, while keeping many individual parts, resources, and tasks connected to the production plan.
Shared operational state
Shipyard ontology
A connected model of the work: what needs to happen, which parts and equipment it needs, who is assigned, and what must finish first.
Illustrative example: forecasting and scheduling repair of a shipyard crane.
OPC-UA telemetry from the crane streams into the ontology; a fault threshold opens a repair task.
The task is linked to the parts, the technicians, and the equipment that must stand in.
Lifts that depend on the crane are linked to it, so the impact is explicit.
OR-Tools CP-SAT re-solves against resource and task constraints; the planner reviews the result.
If a part arrives late or equipment becomes unavailable, these connections show which tasks are affected and give the planning system the information it needs to evaluate a revised schedule.
Implementation and engineering decisions
Model the dependencies
Python/FastAPI services expose an ontology connecting parts, equipment, workers, and tasks. The model gives the scheduling and interface layers a shared representation of operational state.
Make planning constraints explicit
OR-Tools CP-SAT evaluates schedules against task and resource constraints. AI-assisted analysis supports planning without replacing the explicit constraints used by the scheduler.
Preserve accountability
Role-based access, event history, and tamper-evident audit records preserve operational changes and recommendations for review. Docker packages the services for local deployment.