Some clients bring me a metering platform that needs to scale. Others need someone to sit with their team, make the hard architectural calls, and stay until the thing is in production. Both are the same job to me, and I'd rather hear about yours in a few plain sentences than in a brief.
Write to me at hello@rob.engineering. I answer every mail myself, usually the same day.
Metering, load management, device fleets, the protocols nobody enjoys reading about. More than a decade in this domain, from firmware-adjacent C up through Go services and the cloud they run in. Fewer surprises for you, and a much shorter ramp-up.
For teams that need senior judgement more than another pair of hands. Architecture and build-versus-buy calls, hiring and mentoring, roadmaps that survive contact with reality, and the unglamorous work of getting engineering and the business to agree on what matters.
Describe a problem in enough detail and the solution — and the tools it needs — become obvious. That's where I start: understanding the problem fully, then picking whatever fits it best — the language is rarely the hard part. Lately that's meant a lot of Go and C/C++, from embedded to serverless, and increasingly LLMs and reasoning models where they genuinely earn their keep. Next year it may mean something else.
You describe the situation, I ask blunt questions. If I'm not the right person I'll say so, and I usually know who is.
One or two weeks with a defined outcome — a review, a prototype, an architecture you can act on. Small commitment, real answer.
Fixed days per week, a written update every Friday, everything in your repos and your tools. Cancel with two weeks' notice.
Documented decisions and a team able to carry it. A good engagement ends with you not needing me.
Not terms and conditions — just the handful of things I've decided in advance so neither of us has to negotiate them later.
I don't subcontract. The person on the call is the person doing the work.
No fixed price on work nobody has specified yet. Days and materials, and you can stop whenever it stops being useful.
Your code lives in your repositories from day one. No handover cliff at the end.
Clients stay confidential, including you. Nothing on this page names anyone.
I'll tell you when the honest answer is that you don't need me.
Clients are confidential, so entries are recorded by shape rather than by name. Detail on request, in person.
A residential battery fleet had to prove to a transmission operator that it could deliver balancing power. Drove the prequalification end to end — regulatory analysis, control-path architecture across three organizations including the transmission system operators (4ÜNB), latency measurements on the lab bench — through to the submission evidence.
A solar company with a large installed fleet of home energy systems wanted them to act as one virtual power plant in the energy markets. Drove the architecture work that gets a device fleet ready for that: a redesign of the on-device energy manager to serve self-consumption, § 14a EnWG grid constraints and market control at once, control-loop simulations to test strategies before the fleet ran them, and the cross-team RFCs that carried the decisions.
A storage pilot for apartment buildings: a battery behind every tenant meter, energy shared across the house under rules (§ 42b EnWG) new enough that nobody had built against them yet. Shaped the architecture around what the pilot could actually afford — battery control through the manufacturer's cloud instead of on-site controllers, meter data over the smart-meter-gateway infrastructure — and took on most of the build.
A solar distributor wanted its installers' whole value chain on one platform — from offer and system design through ordering and rollout to maintenance, across photovoltaics, battery storage and EV charging. Wrote the proposal and the architecture record, then built it: a modular Go backend, a React frontend, and a live integration into the distributor's ERP for catalog, stock and ordering.
An energy-software startup building modular blocks that other companies assemble into their own energy products; the entry project is analytics for a solar installer who also operates what they build. Predictive maintenance from the plants' own measurements — fleet-wide monitoring, specific yield (kWh/kWp) benchmarked against forecasts, strings and near-identical systems compared against each other — so the morning after a storm shows where the damage is. Hands-on through the whole stack, from Go services to cluster infrastructure.
The employed years before consulting: led the integration domain at an energy-management company — the gateways in the field and every device family they spoke to. Inverters, batteries, EV chargers and heat pumps over Modbus TCP and RTU, EEBus, OCPP, SG Ready and the manufacturers' cloud APIs; responsibility ran from bootloader and embedded OS up through MQTT, gRPC and REST interfaces to the cloud services behind the fleet — and for the teams that built them.