SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Fluidstack is building civilization-scale compute infrastructure for AI, acquiring power, designing and operating data centers at gigawatt scale. The Facility Controls team owns the real-time control systems that manage these facilities: load control, mechanical/electrical/plumbing (MEP) integration, behind-the-meter systems, automated commissioning, and autonomous deployment.
In this role, you will build and own the facility-side energy control contract—the production services that execute it. This includes the protocol layer, gateway service, and API that third-party plant controllers poll against. You will design and evolve the messaging and state layer that computes the facility's load intent from live telemetry and publishes it every second with forecasts. You own the data contracts and engineering standards that other teams build against, ensuring they remain consistent as the platform scales to new sites and counterparties.
Key responsibilities include building the constraint execution path (answering incoming requests with what the facility can achieve and by when, within second-level deadlines), and building observability systems that prove conformance in production—per-stage latency, staleness, watchdog health—so missed budgets are visible before they become incidents.
The team operates with full autonomy, insane urgency, first-principles reasoning, and a focus on building something that matters. You will work across the full signal chain from device to database, treating data integrity as non-negotiable. You design before you build, draw clear boundaries on failure domains, and move toward production problems with urgency and without drama.
REQUIREMENTS:
- Extensive experience with high-throughput messaging systems (NATS, Kafka, or equivalent) at real scale; deep knowledge of failure modes
- Designed time-series data models in ClickHouse, TimescaleDB, or comparable systems; understand tradeoffs between write throughput, query performance, and schema evolution
- Instrumented production services with observability tooling (Prometheus, Grafana, or equivalent); treat metrics and alerting as part of shipping
- Built and owned production services that other teams depend on; ship code you would be comfortable being paged for at 3am
- Design-first approach: can name patterns you reach for, explain rejected alternatives, and point to deliberate refactors
- Draw clear boundaries on interfaces, modules, and failure domains; write tests that catch failures you actually fear
- Worked across full signal chain from device to database; treat data integrity as non-negotiable; handle stale vs. wrong values differently
- Move toward broken production pipelines with urgency and without drama
BONUS QUALIFICATIONS:
- Experience with Fluidstack's stack: Go, NATS, Redis
- Industrial and utility protocols: DNP3, Modbus TCP, OPC UA
- Kubernetes and ArgoCD for production service deployment
- Real-time control and dispatch systems: SCADA, EMS, power plant controllers
- Real-time dashboarding with Grafana or equivalent
