Forward Deployed Engineer at Fluidstack: Role, Pay, and Interview (2026)
Fluidstack has quietly run two completely different Forward Deployed Engineer jobs under one title in the past year, and which one you apply to changes everything: the work, the pay, and who your “customer” is.
The first version, posted in 2025, was the textbook model: a customer-facing engineer who deploys 1,000+ GPU clusters for AI labs, debugs them, and carries an on-call rotation. The second, posted October 8, 2026, is stranger and more interesting. Three FDE seats sit on an internal “Acceleration” team, and the customers are Fluidstack’s own supply chain, compute operations and construction teams. This is the clearest example on the board of the title drifting from “engineer at the customer” to “engineer with maximum autonomy.”
The company, briefly
Fluidstack builds AI compute infrastructure. Anthropic selected it to help build $50 billion of US data-center capacity, with first facilities in Texas and New York expected online through 2026. On July 20, 2026 it confirmed an $830 million Series A at a $7.5 billion valuation, led by Situational Awareness, larger than the $750M-at-$7B figure reported earlier. Fluidstack says it can compress multi-gigawatt construction to about six months using factory-built prefabricated modules, against several years today.
Revenue and headcount are not public. Sacra estimates roughly $180M ARR at the end of 2024, but that is a third-party estimate, so treat it as directional.
Version one: the customer-facing FDE (2025)
An archived 2025 posting (since removed, mirrored on Built In) describes the classic job: act as the customer’s technical advisor across the lifecycle of their AI workloads, deploy GPU clusters of 1,000+ GPUs with custom playbooks, validate compute, storage and network performance, migrate large datasets off public cloud, and join an on-call rotation of up to two weeks a month.
| Dimension | 2025 posting |
|---|---|
| Experience | 2+ years in SWE, SRE, DevOps, sysadmin or HPC |
| Skills | Kubernetes and/or SLURM; Go, Python or Bash; Ansible or Terraform |
| Preferred | 1,000+ GPU workloads, InfiniBand/RoCE, petabyte-scale storage (Ceph, Lustre, Weka, VAST) |
| Base pay | $100K–$250K plus options |
| Location | In-office, four locations |
That is an infrastructure-operator profile, closer to SRE than to the application-layer roles we cover in Cursor or Cohere.
Version two: the Acceleration team (October 2026)
The three new postings are Compute Operations (New York), Manufacturing & Supply Chain (New York), and Infrastructure Development & Delivery (San Francisco). Only the supply-chain posting publishes its full text; the other two expose just titles and team names when read, so everything below on duties and pay comes from that one posting.
It lists the team as Software & Automation and the pay as $224K to $300K plus equity, onsite in New York. The work is unglamorous and very concrete: generating purchase orders from bill-of-materials templates, building procurement tracking that flags late lines and vendor drift, matching deliveries to PO lines and serial numbers by warehouse bin, and shipping manufacturing software for engineering change notices, inspection plans and nonconformance workflows. You embed with sourcing leads, quality managers and warehouse teams and turn their judgment into systems.
Requirements: production Go, Python or TypeScript; LLM API, MCP or agent-framework experience including document extraction on messy inputs; daily use of Claude Code or Cursor; credibility with factory, warehouse or procurement operators. Bonus points for MES/ERP/WMS knowledge, BOM lineage and supplier quality (NCRs, CAPA).
Our contrarian read: this is not FDE as customer consulting, it is FDE as internal automation under extreme schedule pressure. When your product is compressing an 18-to-24-month build cycle toward six, the bottleneck is procurement paperwork, and Fluidstack is staffing it with software engineers rather than ops analysts. A July 2026 founder post framed an earlier version of the role as making “ordering 10,000 GPUs like opening a PR.” If that sentence excites you, this is your seat. If you want customer exposure, apply to the 2025-style infrastructure roles instead. For the broader pattern, see what is a forward deployed engineer.
Compensation: what’s solid and what isn’t
The one verified figure for the 2026 seats is $224K–$300K plus equity (supply chain). The 2025 posting showed base of $100K–$250K, and our job board’s earlier July entry listed $150K–$275K from a founder hiring post. Those are different roles at different times, not a raise. Compute Operations and Infrastructure Delivery show no pay we could read, so ask early. Benchmark against the bands in our FDE salary guide, and press hard on equity: at a $7.5B private valuation, how many options, what strike, and what refresh cycle matter more than the cash band.
The interview loop
Only the 2025 process is documented: a 15-minute hiring-manager call, then three 45-minute virtual interviews covering a technical deep dive, customer communication and debugging, and culture fit, with a stated aim of finishing in about a week. Fluidstack has not published a loop for the Acceleration roles, so assume it is similar but tilted toward a build exercise. To prepare:
- One ugly-data project. Extraction over vendor emails, PDFs or POs, with an error rate you can quote.
- An operator story. A time you shipped software to people who did not ask for it, and how you drove adoption.
- Speed with a data model. Be ready to explain what you left behind that others could extend.
- Fluency with AI coding tools, shown by example rather than claimed.
Who should apply
Senior generalists who like physical-world systems, and who are comfortable with a flat org and onsite work in New York or San Francisco, will fit. Pure customer-success profiles will not. Fluidstack’s earlier, now-closed listings also appear on our job board, and if you are weighing where the money is, compare the infra-heavy seats against the Anthropic roles it ultimately builds for.
See the open Fluidstack seats on the job board, or get the weekly digest when new ones post.