Energy monitoring

What this was

A platform for optimising electricity costs on a deregulated, US-style market — one where a company can not only buy power but sell it back. The system analysed the market and answered two questions: when it pays to buy, and when it pays to sell. Prices there move hourly, so a decision taken a day late costs money.

Technically this is a high-load problem: a stream of market data has to be ingested continuously, decisions computed on top of it, and nothing lost during peak hours. Dmytro built a defined part of the system and handed it over to the client’s team — that is the honest boundary, and we state it rather than blur it.

What carries over

Metering consumption and producing decisions from a live stream is the same class of problem as industrial telemetry — only instead of a meter on site there is an exchange price. That is why energy work sits next to SCADA and dispatching here: what they share is not the sector but the shape — data arrives continuously, and the decision has to be made while it is still fresh.

Adjacent work already running in production: a network of sites with their own controllers, telemetry that survives hours of lost connectivity and sends the backlog afterwards, and dashboards where the owner sees every device without driving out to it.