Staff Software Engineer, (DEM)

Crusoe•Sunnyvale, CA

2d•$215,000 - $260,000

About The Position

Crusoe is hiring a Staff Software Engineer to help build Dynamic Power Management (DPM), a mission-critical system that measures, forecasts, and orchestrates energy use across data centers—from each site down to individual GPUs. DPM aims to direct power where it produces the most useful compute, improve utilization of available megawatts, keep sites operating smoothly as electrical conditions change, and deliver consistent customer performance. As a founding engineer, you will work with the principal architect to translate product requirements into a production architecture, ship the first version, and help shape the growing team. The work spans GPU energy management and hardware telemetry through capacity planning and workload placement, with a direct connection to revenue. The system must be precise, observable, dependable, and safe because it operates beneath customer workloads. You will build the power-control loop, telemetry pipelines, power models and forecasts, workload energy profiling, and simulation and trace-replay tools. You will guide the system from shadow operation toward automated production management, define safety and failure behavior, connect DPM to capacity planning and VM placement, establish success metrics, and lead cross-functional design work with engineering, energy, facility, capacity-planning, and hardware teams.

Requirements

10+ years of industry experience, or 8+ years with clear technical leadership, building and operating distributed systems in production.
Experience leading a system from design through launch.
Experience building control systems or closed-loop automation where incorrect outputs have physical or financial consequences.
Experience turning forecasting or simulation work into production services on which other systems depend.
Depth with high-volume time-series telemetry, including ingestion, aggregation, and reasoning about staleness, gaps, and measurement error.
Fluency in Go or Rust.
Comfort working directly with hardware and facility teams, and ability to read or quickly learn to read an electrical one-line diagram.
Ability to explain technical risk to senior engineering and business leaders in terms of megawatts, dollars, and customer impact, and influence decisions.
Strong written design skills.
Commitment to Crusoe’s company values.

Nice To Haves

Experience with GPU energy management, such as NVIDIA DCGM/NVML, NVIDIA DPS, or AMD equivalents.
Experience integrating with BMCs (Redfish), PDUs, building management (BMS), or electrical power monitoring (EPMS) systems.
Experience operating large GPU or HPC fleets, or working in data center energy or capacity planning.
Experience with optimization and probabilistic methods for allocating scarce resources under risk, such as constraint solvers or failure analysis.
Experience putting a dollar value on infrastructure choices, such as stranded power, build-out timing, or hardware utilization.
Experience with energy optimization software for generation, battery storage, or large grid-connected sites.
Experience with workload-aware scheduling using Kubernetes or Slurm.
Experience with predictive scaling or load forecasting at scale.

Responsibilities

Design and build the DPM control loop to allocate power across sites, rows, racks, nodes, and GPUs, apply decisions through GPU and chassis power management, and adapt as conditions change.
Own high-frequency power telemetry pipelines for GPUs, NICs, chassis and BMCs, PDUs, and facility systems, accounting for latency, accuracy, stale data, and missing data.
Develop power models and forecasts for site, row, and rack needs, idle capacity, and uncertainty using live telemetry and simulation.
Profile the energy use of training and inference workloads, identify power use that does not improve results, and develop efficiency policies.
Build tools to replay production power and workload traces and validate policy changes offline before production use.
Roll out DPM through shadow operation, comparing production recommendations with actual usage before enabling automated management.
Define fail-safe defaults, behavior for stale telemetry, limits on the speed of setting changes, audit trails, and verification that adjustments had the intended effect.
Connect DPM to capacity planning and VM placement.
Define and communicate success metrics and tradeoffs, including compute delivered per megawatt and performance per watt.
Collaborate with data center electrical and mechanical engineering, the energy team, capacity planning, and hardware vendors on interfaces, constraints, and roadmaps.
Lead design reviews and mentor engineers across teams.
Work with the principal architect to turn product requirements into a production architecture, ship the first version, and help shape the team.
Write strong technical designs and RFCs for other engineers to build from.