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, with a mission to deploy 10–100s of GW of compute faster than anyone else. The company is rethinking every layer of the stack, from power acquisition through data center design and operation, with teams spanning hardware and software.
The Thermofluids R&D team is tackling unprecedented thermal challenges: rejecting more than 10 gigawatts of heat starting in 2027, designing and building custom cooling machines (compressors, chillers) that the merchant market cannot supply at scale, and owning the full thermofluids problem across the fleet—working fluid selection, cycle architecture, heat exchangers, two-phase heat transfer, and thermal energy storage.
In this role, you will:
- Define the thermodynamic cycle model that sets working fluid, pressures, temperatures, and flows for compressors, turbines, and heat exchangers.
- Design heat exchangers with surface areas in the millions of square meters, among the largest cost drivers of a data center's footprint.
- Design the fluid system (piping, valves, headers, relief, isolation) sized against pressure-drop budgets and actual operating pressures.
- Run the trades that determine machine performance: air-side temperature rise, face velocity, subcooling. Derive operating points from coupled exchanger ratings to catch infeasible designs in the model, not in hardware.
- Take cycle predictions to the test stand, reconcile them against measured COP, and report honestly where the model diverged from reality.
The team operates with full autonomy, insane urgency, first-principles reasoning, and a focus on building something that matters. You will own problems end-to-end and drive everything forward as fast as possible.
REQUIREMENTS:
- You have built a thermodynamic cycle model from fluid properties up and defended its assumptions to someone actively trying to break it.
- You have sized real heat exchangers against published correlations and caught cases where the correlation belonged to a different geometry.
- You have calibrated a model against machine datasheets or test data and reported gaps honestly rather than tuning them away.
- You have written analysis that changed a design decision, in a form that non-programmers could read and act on.
- You have sized a relief path and signed off that the system was protected.
- You have held a result open when it was still open, instead of closing it early to be helpful.
BONUS EXPERIENCE:
- Vapor-compression and transcritical cycles (chillers, heat pumps, refrigeration).
- Propulsion and cryogenic fluid systems (feed systems, regenerative cooling, two-phase transport).
- Piping and relief design for high-pressure fluid systems.
- Real-gas properties and two-phase heat-transfer correlations (CoolProp, REFPROP, Cavallini, Gnielinski, Wang).
- Data center thermal architecture (CDUs, dry coolers, direct-to-chip).
