The hiring profile
In March 2026, Travis Kalanick's City Storage Systems rebranded as Atoms and took a $1.7 billion equity investment from Andreessen Horowitz to build a shared wheelbase for specialized industrial machines that go into mines, food production lines, and transport networks. The company runs a team of 1,000 to 5,000 across more than 110 operating cities. Its careers page lists 97 open roles in four buckets: Engineering & Research & Product (34), Operations (24), GTM & Customer Success (6), and a catch-all "Other" (33) covering finance, legal, recruiting, and internal systems. Software Engineering leads with 17 postings; Finance & Accounting follows with 9. The long tail (HR, Recruiting, Design, ML/AI, Product, Marketing, Sales, Infrastructure, Analytics, Legal, Data Engineering, QA, IT) plus 33 unclassified roles signals an org chart still being drawn.
The hiring footprint spans Austin, San Francisco, Los Angeles, New York, and Riyadh. Every office-based role requires five-day onsite presence. That spread mirrors the operational reality: Atoms deploys machines into physical sites, not cloud regions, so the people who integrate sensors, compute, manufacturing, and real estate need to be where the equipment lives. The company's own language — "cross-stack competence wins," "high AI fluency," "comfortable operating as a generalist" — describes a profile that blends robotics, embedded software, fleet operations, and the grit to debug a machine on a mine site at 2 a.m.
Early-career talent gets a dedicated pipeline. The site flags an intern program for students, recent graduates, and researchers who want to apply AI and robotics to the physical world; interns contribute to live products. Senior roles confirm the company is staffing functional leadership alongside the technical core: Vice President of Global Controller, Vice President of Finance Technology, Staff Machine Learning Engineer. A product manager for integration and ecosystem sits beside a construction project manager and a commercial HVAC/R technician, all posted the same week. The "progress machines" framing means the bar selects for people who have shipped something that touches atoms, not just bits, and who can operate from sensor calibration to fleet telemetry to site-level economics.
What the pay looks like
Atoms' compensation splits into two tiers: an executive-and-specialist band at $224,000–$360,000, as Zero G Talent's recent postings show, and a broader engineering-and-operations band at $62,000–$284,000, as Zero G Talent's board summary puts it, with a $190,000 median across 71 salaried roles. The Zero G Talent board — live listings ingested directly from the source — provides the most reliable yardstick.
| Role (location) | Salary band (USD/year) | Source |
|---|---|---|
| Vice President, Global Controller (SF / NY / LA) | $300,000 – $360,000 | Zero G Talent board |
| Vice President of Finance Technology (SF) | $300,000 – $350,000 | Zero G Talent board |
| Staff Machine Learning Engineer (SF) | $273,000 – $345,000 | Zero G Talent board |
| Infrastructure Product Manager (SF) | $224,000 – $284,000 | Zero G Talent board |
All four VP-level slots sit at the top of the range; the staff ML engineer and product manager roles anchor the next tier down. A LinkedIn posting for a Software Engineer in Money Engineering (New York, July 2026) quotes a base of $145,000–$183,250 plus equity awards and an annual performance bonus, with no cash bonus guaranteed. That posting also noted "Over 200 applicants" and that referrals double interview odds. Treat the LinkedIn band as the current floor for a mid-level engineering hire in New York; the board's $190,000 median suggests most technical hires land above it.
The LinkedIn posting confirms "equity awards" but does not disclose grant sizes, strike prices, or refresh policies. BuiltIn's 2026 review flags "unclear equity availability" and "opacity around compensation specifics" as weak points, noting role-dependent variability that warrants direct confirmation. Contractors should note BuiltIn's callout of "uneven benefits access for contractors"; the full suite below applies to employees.
Benefits are substantial for a small team. The LinkedIn posting lists medical, dental, and vision insurance with multiple plans including HSA options; company-paid life and disability insurance, short- and long-term; voluntary accident, critical illness, and hospital indemnity coverage; optional supplemental life insurance for self, spouse, and children; a pet insurance discount; 401(k); Health Savings Account; Flexible Spending Accounts for healthcare, dependent care, and commuter expenses; discretionary vacation days; eight paid holidays; paid sick time; paid bereavement leave; and paid parental leave. BuiltIn's 2026 assessment calls out strengths in healthcare coverage, retirement availability, and a broadly described leave set, while noting the explicit five-day in-office model. The posting adds a standard disclaimer: benefits are subject to change at the company's discretion.
Inside the interview loop
Google found four interviews the sweet spot; beyond that, predictive power flatlined. Korn Ferry's six-stage model maps the industry standard: a background-and-fit screen for basic qualifications, location flexibility, and salary alignment; a competency-and-behavior stage with open-ended prompts; a team round where future peers assess capability and daily-work compatibility; a work-product demand (a presentation, a go-to-market plan, a case study); a skip-level meeting with a senior leader that signals serious intent; and negotiation.
What separates strong processes from long ones? A YouTube deep-dive on engineering hiring (December 2025) argued that effective loops explicitly test on-the-spot reasoning about tradeoffs, not recall or rehearsed stories. A well-designed session might hand a candidate a five-page postmortem, give five minutes to absorb it, then debate whether the team structure should change based on conflicting evidence buried three pages apart. That task demands working-memory capacity, domain fluency, and the ability to synthesize under time pressure, things chatbots and cramming can't fake. IQ tests, while controversial, appear in screening stages at high-standard firms precisely because they're unpracticable and filter quickly, saving both sides time.
Countermeasures that raise signal: panel interviews so every evaluator hears identical evidence; work assignments (LinkedIn calls them the most effective yet underused technique); standardized evaluation forms that force comparison on fixed criteria rather than gut feel; interviewer training that includes implicit-bias awareness and role-play calibration. Transparency helps too: publishing the process online, sharing the STAR answer framework upfront, and committing to a decision timeline (ideally within a week) all correlate with higher offer-acceptance rates and better candidate experience.
Atoms' specific interview architecture isn't documented in available sources. Candidates should prepare for a structured, multi-stage loop that stresses job-relevant problem solving over trivia, expect a work-sample or case component, and treat every conversation (especially the team round) as a two-way assessment. Processes that feel rigorous but fair, transparent, and time-bound land the people who stay. The same discipline that shapes Atoms' machine deployments — instrument, iterate, ship — applies to how it evaluates the people who build them.
Where the machines live
Atoms operates from a Los Angeles headquarters at 633 W 5th Street, a downtown address anchoring three divisions: Atoms Food, Atoms Mining, and Atoms Transport. The headquarters sits in a city that has become a quiet hub for hardware-heavy automation, close to port logistics, aerospace supply chains, and a labor pool spanning robotics, manufacturing, and real-estate development. But the work does not stay in one building.
Active roles in San Francisco and New York signal a bicoastal footprint for finance leadership, core ML, and platform product. A Vice President of Global Controller search ran simultaneously in all three cities; a Staff Machine Learning Engineer role and an Infrastructure Product Manager role both listed San Francisco. In May 2026, Atoms hosted a builder gathering in San Francisco for teams working across robotics, software, operations, and the physical AI stack, an event that doubles as a recruiting signal and confirms the Bay Area office as a technical center of gravity, not just a sales outpost.
Atoms Food is the largest operational surface area. It comprises CloudKitchens (delivery-first kitchen infrastructure), Otter (restaurant operating system), Lab37 (robotics and automation for food production), Picnic (office lunch logistics), and ProFood Properties (facilities built for modern food brands). CloudKitchens and ProFood Properties develop and operate kitchen real estate, essentially distributed micro-factories for food. Lab37 is the explicit robotics arm: its mandate is automation for food production. Picnic adds a last-mile logistics layer touching office buildings and dense urban routes. Together, these businesses give Atoms a network of kitchen sites, fulfillment nodes, and robotics test beds that span multiple metros, a distributed lab no single campus could replicate.
Atoms Mining runs through Pronto, the off-road autonomy company Atoms acquired in May 2026. Pronto's technology has already been proven at industrial scale in surface mining. The test cells for Atoms Mining are not in a lab; they are haul roads, loading zones, and pit operations at active mine sites. The acquisition announcement framed the next step as "faster innovation and a technology roadmap that will extend far beyond what any standalone company could deliver," backed by Atoms' "unmatched resources of a polymath organization that has spent nearly a decade building at the intersection of AI, robotics, sensors, manufacturing, and real estate." In practice, that translates to mine-site deployments where perception stacks, planning, and fleet coordination run on production ore bodies, not proving grounds. The feedback loop is measured in tonnes moved and uptime hours, not benchmark scores.
Atoms Transport — described as "wheelbase for robots" — suggests a hardware platform team building the mobile base layer other divisions can standardize on. That work demands a machine shop, dyno or rolling-road test capability, and environmental chambers for vibration, thermal, and ingress validation. No dedicated Transport facility is named, but the division's existence implies a hardware prototyping and validation shop, likely co-located with or near the LA headquarters where manufacturing supply chains are accessible.
The $1.7 billion equity investment led by Andreessen Horowitz, announced in July 2026, merged the operating businesses into a single Atoms equity structure and was explicitly labeled as "fuel for the next phase: more machines, more deployments, more of the hardest problems in the physical world attacked at once." Capital at that scale in a privately held automation company typically funds three things: new facility build-out or long-term leases on industrial space, capital equipment for in-house fabrication and test, and working capital to staff multi-shift operations at deployed sites. The company's own language ("gainfully employed robots," "specialized, purpose-built progress machines designed for high-cycle industrial environments") points to a strategy of owning the hardware loop: design, build, deploy, instrument, iterate.
What emerges is not a campus model but a constellation: a downtown LA anchor for leadership, systems engineering, and cross-cutting platform work; a San Francisco node for core AI/ML and product; a New York node for finance and possibly capital-markets-facing functions; a distributed network of CloudKitchens and ProFood Properties sites that double as production test beds for Lab37 robotics; active mine sites where Pronto's autonomy stack runs on customer ore; and wherever the Transport wheelbase is being fabricated and shaken down. The common thread: every site is a revenue-generating or data-generating asset; there are no pure research labs divorced from deployment. The test cell is the kitchen line, the mine haul road, the delivery route. The machine shop where the next progress machine is assembled for its first shift is the environment Atoms hires for.
Who lasts
The signals Atoms projects (through product messaging, portfolio companies, and the roles it prices at the top of market) form a coherent picture of the profile that succeeds inside. None of this comes from a published values page or employee testimonials; the research contains neither. What exists is a pattern of emphasis across atoms.co that describes the operating DNA the company selects for.
Technical depth is non-negotiable at the senior level. The board data shows a Staff Machine Learning Engineer banded at $273,000–$345,000 in San Francisco. The portfolio companies (CloudKitchens, Pronto AI, Lab37, Otter, Picnic, ProFood) each operate as vertically integrated units in food, mining, or robotics where the founding team owns the full stack.
Cross-functional fluency is priced into the leadership roles. The board lists Vice President, Global Controller and Vice President of Finance Technology both at $300,000–$360,000, alongside the Staff ML Engineer and an Infrastructure Product Manager at $224,000–$284,000. That clustering suggests the company expects finance, infrastructure, and product to operate as peers, not as support functions. A controller who cannot speak infrastructure, or a product manager who treats finance as a black box, will struggle in the collaboration model the compensation structure encodes.
The people who last are the ones who would rather be there than anywhere else.
Working in frontier tech? Zero G Talent tracks the openings: see every open Atoms role, browse frontier tech jobs, the companies hiring, and the people building the field.