Building for the factory floor

The context

SME Factory Insight is an industrial IoT and manufacturing intelligence product for factory clients. Owners, supervisors, and operators need different views of the same operation: machine status, production, shifts, downtime, OEE, and energy use.

My work connects those needs to web and mobile features, backend services, telemetry integrations, and production delivery.

My contribution

I develop React and TypeScript dashboards and React Native / Expo mobile workflows, alongside Node.js APIs and PostgreSQL-backed services. The product work includes production and shift reporting, downtime reasons, product counts, maintenance, and energy monitoring.

This role also involves understanding client workflows and turning them into usable software. A reporting feature needs to fit how people work on the floor, as well as how the data is stored.

The migration

I handled a live migration of approximately four million rows while moving the product from Lovable to a custom VPS, preserving existing data. Infrastructure work became part of delivering the product, alongside ongoing feature development.

My responsibilities also span Linux VPS deployment, Docker, PM2, Nginx, monitoring, backups, and release / rollback workflows.

Keeping telemetry useful

I worked on buffered machine-data ingestion using MQTT and Redis, with replay into PostgreSQL. Incoming data is buffered so a database or backend interruption does not immediately block collection; replay retains the original timestamps.

The distinction matters for manufacturing reports: when a reading happened is different from when a backend received it. Production and shift views need to respect the event time.

Machine telemetry -> MQTT -> Redis buffer
                              |
                              v
                     PostgreSQL + backend
                              |
                              v
                   Web and mobile workflows

The result

The migration preserved existing data while moving approximately four million rows to the custom VPS. My wider contribution includes delivering product features for end clients and taking responsibility across implementation and production operations.

The product continues to evolve. Work such as owner-facing AI insights has its own release gates; a capability in development is not automatically a generally available feature.

What I take forward

I treat infrastructure and product usability as connected work. A clear dashboard depends on reliable data, understandable workflows, and a deployment that can be operated.

For future migrations, the questions I want answered explicitly are what data is in scope, how success will be checked, what recovery looks like, and who owns the product after the move.