# Starcloud interview guide for space-compute engineers

Source: https://www.techinterview.org/post/3233477958/starcloud-interview-guide/
Updated: 2026-11-16 · techinterview.org

There is no public candidate record of the Starcloud interview process yet, so treat any round-by-round sequence you read here as inferred, not reported. What is solid and sourced: Starcloud is a Woodinville, Washington startup (founded 2024, formerly Lumen Orbit) building solar-powered, GPU-carrying data centers in orbit. It raised a $170M Series A at a $1.1B valuation in March 2026 and a $250M extension at a roughly $2.3B valuation in August 2026, and in November 2025 it flew an Nvidia H100 on its Starcloud-1 satellite, which Nvidia called the first data-center-class GPU in orbit. For an early company hiring mechanical, thermal, GNC, power-electronics, avionics, and software engineers, a plausible loop is a recruiter or founder screen, a hiring-manager technical round, a deeper technical panel, and a founder round, but that shape is inferred from the open roles, so confirm the real sequence with your recruiter. One catch to know up front: every Starcloud role detail page requires applicants to be a US citizen, permanent resident, or otherwise eligible under US export regulations.

Starcloud is trying to build something that was not a product category two years ago: a data center that runs in orbit, powered by sunlight and cooled by radiating heat into space, with the compute riding on the satellite itself. The pitch is that power and cooling, the two binding constraints on terrestrial AI data centers, are both cheaper above the atmosphere, where sunlight is near-constant and nobody is bidding against you for grid capacity. Whether that economics holds at scale is the open question the company exists to answer.

The funding says investors want to find out. Starcloud operated as Lumen Orbit until it rebranded in early 2025 after a trademark conflict with the fiber carrier Lumen Technologies ([GeekWire, March 2025](https://www.geekwire.com/2025/lumen-orbit-starcloud-10m-space-data-centers/)). It then raised a $170M Series A at a $1.1B valuation co-led by Benchmark and EQT Ventures, announced in [late March 2026](https://dealroom.co/news/127100-starcloud-raises-170m-to-build-ai-data-centres-in-orbit-hits-1-1b-valuat/), and a $250M Series A extension in [August 2026](https://www.unite.ai/starcloud-raises-250m-series-a-extension-at-2-3b-valuation/) pushed the post-money valuation to about $2.3B. In November 2025 it flew an Nvidia H100 on its Starcloud-1 satellite, which Starcloud [says](https://www.datacenterfrontier.com/site-selection/article/55337494/starcloud-launches-orbital-ai-data-center-with-nvidia-h100-gpu) is the first data-center-class GPU in orbit. For an engineer, the read is a well-capitalized company that is an aerospace hardware shop and a compute-infrastructure startup at once, which is why its hiring spans two worlds that rarely overlap.

## Who Starcloud is hiring, and why it reads like a satellite company

Starcloud lists its open roles on its own careers board at starcloud.octbr.ai, and as of November 2026 every one of the sixteen is in Woodinville, Washington, onsite. The mix leans hard toward spacecraft hardware: mechanical (satellite bus, mechanisms, solar, loads and dynamics), thermal, guidance-navigation-and-control (GNC), power electronics, avionics and harness work, plus a single all-levels software engineering role, internships, and a facilities lead. The center of gravity is closer to a satellite company than a SaaS one: a thermal engineer here is solving radiative cooling for a kilowatt-class GPU in vacuum, not sizing a CDN. The nearest comparisons on this site are a defense-hardware team like [Mach Industries](/companies/mach-industries-interview-guide/), where physical testing and margins drive the interview, and the GPU-infrastructure thinking you would see at a company like [Modal](/companies/modal/). A background in one of those two halves will likely have to show it can reason into the other.

## The Starcloud interview process is not public yet

No candidate has posted a Starcloud or Lumen Orbit interview account on Glassdoor, Blind, Reddit, or YC's own job board that a November 2026 search could find, and the postings on starcloud.octbr.ai describe qualifications, not stages. So everything in this section is inferred from the roles on the board and from how comparable early-stage space-hardware teams tend to run their loops. None of it is a reported sequence. Treat your recruiter's agenda as the source of truth and ask for it in writing before you prep.

A small hardware company usually opens with a recruiter or founder call covering motivation, availability, work authorization, and rough level. A hardware candidate would typically then meet the hiring manager for a technical conversation on their discipline, walking through a design they owned and defending the trade-offs. After that, early-stage aerospace teams tend to run a deeper technical panel, sometimes a short exercise or a review of past work, and at this size a founder or executive conversation is likely near the end rather than reserved for leadership roles. Round counts, durations, and take-home format are all unknown, so prep the content, not the schedule.

## What each track probably screens for

For the aerospace and hardware disciplines, the reported requirement from the postings is a BS in electrical, computer, mechanical, or aerospace engineering (or equivalent experience) plus discipline-specific depth. What an interview would test on top of that is inferred, but the pattern across spacecraft teams is consistent: first-principles reasoning about margins, thermal budgets, vibration and loads, radiation tolerance, and how you design and run a test campaign when a bug in orbit is unrecoverable. A project where you owned a physical system end to end and can speak to what went wrong is the kind of story these rounds tend to look for.

For software, the board carries one all-levels software engineering role and does not publish a stack, so the specifics here are inferred. The work plausibly splits into flight and embedded software on the satellite, ground systems and mission operations, and the data path that moves results from an orbiting GPU back to a customer. Be ready for systems-level design under hard constraints, which is why the multi-tenant, high-ingest problems in the [system design interview guides](/system-design-interview-guides/) are better preparation than pure algorithm drilling, along with the reliability mindset that embedded and ground software demand. Treat any claim about a specific language or framework as unconfirmed until the recruiter names it.

| Starcloud interview and company facts (as of November 2026) | Detail | How well-sourced |
| --- | --- | --- |
| What it does | Woodinville, Washington startup building solar-powered data centers in orbit, with GPUs on the satellite. Founded 2024; operated as Lumen Orbit until an early-2025 rebrand after a trademark conflict with Lumen Technologies | Reported (company site, GeekWire, DataCenterDynamics) |
| Funding | $170M Series A at a $1.1B valuation co-led by Benchmark and EQT Ventures, announced late March 2026; a $250M Series A extension in August 2026 raised the post-money valuation to about $2.3B. Cumulative total reported around $450M since 2024 (figures vary slightly by outlet) | Reported (dealroom, unite.ai, VC trade press); exact cumulative total varies by source |
| Hardware milestone | Starcloud-1, a ~60 kg satellite, launched November 2025 on SpaceX's Bandwagon-4 rideshare carrying an Nvidia H100; Starcloud says it is the first data-center-class GPU in orbit. Follow-on Starcloud-2 reportedly slipped toward 2027 | Reported (space and data-center press); "first in orbit" is Starcloud's own characterization |
| Open roles | 16 roles on Starcloud's own board (starcloud.octbr.ai), all in Woodinville, WA, onsite: mechanical (bus, mechanisms, solar, loads/dynamics), thermal, GNC, power electronics, avionics/harness, one all-levels software engineer, internships, a facilities lead, and a TS/SCI-cleared security manager. Example qualification: BS in electrical, computer, mechanical, or aerospace engineering, or equivalent | Reported (Starcloud careers board, read November 2026) |
| Interview loop | No published loop and no candidate reports under either name. Plausible shape for an early company: recruiter/founder screen, hiring-manager technical round on your discipline, a deeper technical panel or work review, and a founder conversation. Round count, durations, and take-home format all unknown | Inferred from the open roles and comparable space-hardware teams; not reported. Confirm with the recruiter |
| Compensation (base, Woodinville, WA) | Bands appear on each role's detail page (not the index) under WA pay-transparency law: Electrical Engineer (All Levels) $100,000-$200,000; Mechanical Engineer, Mechanisms (All Levels) $105,000-$215,000; a now-removed TS/SCI Senior TPM at $140,000-$240,000. Ranges are wide because each spans all levels; equity not disclosed | Reported from Starcloud role detail pages (read November 2026) |
| Work authorization / export control | Every role detail page requires applicants to be a US citizen, lawful permanent resident, or otherwise eligible under US export regulations. One role also requires an active TS/SCI clearance. Visa sponsorship is not addressed on the postings | Reported from Starcloud role detail pages; confirm sponsorship and your role's requirement with the recruiter |

## Pay, equity, and the work-authorization catch

Anchor compensation to the one source that returns real numbers for a company this new and small: Starcloud's own role detail pages, which must publish a pay band under Washington pay-transparency law. The listing index shows no bands; they sit on each role's detail page. The Electrical Engineer (All Levels) page showed $100,000 to $200,000 and the Mechanical Engineer, Mechanisms (All Levels) page showed $105,000 to $215,000; a now-removed Senior Technical Program Manager role carrying a TS/SCI clearance was posted higher, at $140,000 to $240,000. Those are reported requisition bands, wide because each spans all levels, so ask the recruiter for the band on your exact level. A levels.fyi page will be near-empty for a team this early, so do not rely on it. Equity is where most of the upside sits at a Series A company, and a $2.3B valuation means your options carry real potential and real risk, so run the whole package through a [total-comp calculator](/total-comp-calculator/) before comparing it to a big-tech base; once you have real numbers, the usual [salary-negotiation](/post/3233474669/salary-negotiation-2026/) moves apply.

Work authorization is not a footnote here, and it is the thing most likely to end a non-US applicant's process before it starts. Every Starcloud role detail page read in November 2026 carried the same line: applicants must be a US citizen, lawful permanent resident, or otherwise eligible for authorization under US export regulations, which is the standard export-control gate for space hardware. The board also lists a Senior Information System Security Manager requiring an active TS/SCI clearance, a role that in practice needs US citizenship and signals classified-adjacent work. If you need visa sponsorship, treat this as a likely dead end until proven otherwise: confirm with the recruiter both whether a company this size sponsors work visas at all and whether your specific role clears the export-control requirement, before you invest in the loop.

## How to prep when the public record is this thin

Prep to the problem, since you cannot prep to a known loop. Be able to explain, in plain terms, why orbit changes the compute economics and what makes it hard: constant solar power and radiative cooling on one side, launch mass, radiation, thermal cycling, and unrecoverable failures on the other. A candidate who can argue where this bet breaks will read as a peer; one who only repeats the pitch will not. Beyond that, prep to your track: owned designs and test plans for hardware, system design under physical limits for software.

Because founders are close to every hire at an early company, the behavioral signal carries weight, so bring concrete ownership stories rather than improvising; a [structured STAR answer](/post/3233460379/behavioral-interview-questions-2026-star-method-amazon-leadership-principles-and-winning-answers/) holds up better than a wandering one. For calibration on how early, fast-moving companies run their loops, the [AI-native company interview guides](/ai-startup-interview-guides/) and the [AI-startup interview difficulty index](/ai-startup-interview-difficulty-index/) give more grounded signal than Starcloud's nonexistent review record, and the general [AI-era interviewing](/ai-era-interview-guide/) advice about judgment over enthusiasm applies doubly to a company inventing its own category. The single best move before any Starcloud loop is a blunt email to your recruiter asking for the exact rounds, the format of each, and whether your role carries the export-control or clearance requirement. You will learn more from that reply than from any guide, this one included.
