Engineering
I build the systems underneath. Full-stack architectures, real-time data pipelines, WebGL renderers, spatial APIs. I've led technically and shipped pragmatically – and learned that good architecture is mostly about knowing what not to build.
I’ve built full-stack systems for industrial contexts where real-time isn’t a nice-to-have: WebGL renderers for 3D machine visualization, MQTT/OPC-UA pipelines feeding InfluxDB time-series stores, TypeScript/Node.js backends, React frontends for complex operator interfaces. At JENII I was the sole architect on a greenfield Digital Twin platform – every decision, from data schema to rendering pipeline to deployment, was mine to own. I’ve also led technically: setting standards, reviewing code, mentoring, and knowing when to stop building and start shipping.
APIs have run through all of it: at Paztec, a back-office and API behind every client app, consumed by its mobile front-end; at JENII, the API that made the digital twin possible, written from scratch, alongside a Grafana observability layer and config-export tooling other twins could consume. At Backyou I led the team behind the PMS-integration API — there as the lead, not the coder.
Most recently I’ve been building hands-on with local, open-weight LLMs and agentic coding workflows — on my own hardware, partly out of a conviction that AI worth keeping has to be sustainable, not just capable. It’s exploration rather than production, but it’s the direction I’d take engineering tooling: agents that work inside real workflows, not beside them.