How Work Actually Gets Done at Atomic Industries
The factory floor and the codebase share the same roof. That arrangement, rare in modern manufacturing, dictates the rhythm of every workday at Atomic Industries.
Aaron Slodov founded the company in 2019 after a year calling roughly 200 machine shops nationwide, posing as a business student to learn how the industry actually operated. That research produced a deliberate choice: vertical integration from quoting through final quality inspection, all governed by a single software backbone. The team of roughly 60, including computational physicists, veteran toolmakers, and engineers from SpaceX, Google, and Detroit's manufacturing base, works from three offices with headquarters in Miami and significant operations in Detroit. No separate "software team" receives requirements from a "hardware team." Simulation engineers modeling molten plastic flow and thermal behavior sit beside operators running 1,000-to-2,000-ton presses and Velo3D Sapphire metal 3D printers producing conformal cooling channels.
Decision-making reflects that proximity. When a mold design changes, the simulation updates, the toolpath adjusts, and the press operator sees new parameters without a ticket crossing organizational boundaries. Published lead times (40–60% faster than the traditional 8–16 weeks) and a claimed $0 tooling cost versus the $50,000-plus industry standard are not marketing metrics; they are the daily scoreboard the team optimizes against. That arrangement produces a founder-led engineering culture that moves at rare velocity — and demands a pace that filters for self-starting builders who operate without supervision.
The pace is set by a stated mission to "move customers from design to production faster than anyone on earth" and a founder's framing that reindustrialization requires "a generational amount of work" and "endurance." That language appears in the same breath as descriptions of real-time adaptive control systems adjusting molding parameters across millions of cycles. The work is not abstract optimization — it is physics-first engineering where a simulation error shows up as scrap metal or a missed delivery window within days.
Organizational flatness follows from headcount and integration. With 60 people spanning physics, software, machining, and production, there is no room for middle-management layers. The job board shows open roles for simulation engineers, senior software engineers, and manufacturing engineers alongside finance and purchasing coordinators; a mix confirming the breadth of disciplines operating in a single loop. Salary bands ($79,000–$212,000, median $160,000) reflect the premium placed on engineers who can operate across that loop without handoffs.
| Role | Salary Range |
|---|---|
| Accounting Manager | $130,000–$160,000 |
| Purchasing & Finance Coordinator | $75,000–$85,000 |
| General Engineering Applicant | $95,000–$225,000 (Zero G Talent's board data) |
What a candidate should picture: a day that might start reviewing simulation output from an overnight run, walk to the press floor to validate a mold trial, and end writing the automation that closes the gap between the two. The tempo is set by the slowest physical cycle (molten plastic cooling in a mold) while the software cycle runs at commit speed. The friction between those clocks is where the work lives.
Values and Operating Principles
Atomic Industries does not publish a values page. Its operating principles emerge from what its founders say in interviews, how they describe the problems they chose to solve, and the architecture of the product they built. The through-line is a single argument: the United States lost the ability to make things fast because it let the knowledge of how to make them atrophy, and the only way to recover that speed is to embed the missing knowledge into software that runs the factory floor.
Decision speed compounds
The clearest stated principle comes from board member Jon McNeill, former Tesla president, describing a lesson he took directly from Elon Musk: "the thing that will separate us from all of our competitors is decision speed, because decision speed compounds. Like, I make a decision today, I build on that decision tomorrow, et cetera, and it takes one of our competitors, like Ford or Toyota, 30 days to make the first decision." McNeill said Atomic's product evolved from an optimization platform that gave recommendations to one that "not only gives recommendations but it makes decisions, so it's fully autonomous in that sense." DoorDash, a customer McNeill describes as "DoorDash-scale," runs roughly 90 percent of its purchasing across hundreds of sites through that autonomous layer. The principle is not abstract — it is the product's output.
Techno-industrialism: bits serve atoms
Slodov defines "techno-industrialism" as the merging of two historically distant fields: advanced technology and heavy industry. He argues that decades of offshoring hollowed out U.S. manufacturing and created a "dangerous loss of trade knowledge." The pandemic and supply-chain crises exposed the vulnerability. Atomic's response is to build software that does not sit above the plant but sits inside the operating loop. This is what CVector's Richard Zhang, working on a parallel problem, calls "operational economics": "We position it to sit between the operation of the plant and the actual economics: the margin of how much you're making money." For Atomic, that means software that takes a CAD file and produces a production-ready mold in hours instead of months, closing the loop between design intent and physical reality.
Digitizing tribal knowledge, not replacing it
The origin story is specific. Slodov found a trade built on "wizard-like tribal knowledge bearing trades people" where expertise transfers only through "repeating or going through the same process like hundreds or thousands of times over decades." The crisis moment: a customer put out a public request to rebuild a mold for a B2 part because "the people who originally designed it [were] gone. They don't have like the drawings laying around... It's literally the whole like we forgot how to build it situation." Atomic's approach draws from the self-driving playbook: real-time data from CNC machines and 3D printers trains software to replicate the judgment of veteran toolmakers. Slodov is explicit: "not to just like automate it and get rid of the trade obviously, but just like evolve it into the 21st century because we don't have time to sit around and train like hundreds of thousands of people."
First-principles architecture with software at the core
Slodov compares Atomic's model to Tesla and SpaceX: "all of those production lines are software oriented. They're like their entire process is kind of software oriented." The operating principle is that you cannot bolt AI onto a fragmented process; you must "architect the entire operation from first principles with software at the core." That means vertical integration of design, simulation, and production under one roof, collapsing the slow, fragmented handoffs that define traditional mass production. The flywheel effect Slodov describes is deliberate: "direct application of the technology trying to like collapse um you know process or human labor or engineering down so that like you have more of a flywheel effect of how quickly something happens."
Exoscaling the industrial base
The stated mission borrows a computing term: "exoscale the industrial base" by dramatically accelerating how quickly designs move into production. Slodov frames this as a generational project — "rebuild U.S. industrial strength, creating not just efficiency but renewed prestige and purpose in making the physical things society depends on." He cites Japan's monozukuri (craftsmanship in making) and Ohio's historical industrial rise as cultural precedents. The New American Industrial Alliance (NAIA), which Slodov helped form, extends the principle into policy: using technology developed over recent decades as the foundation for rebuilding the American industrial base.
Venture constraints as a forcing function
Slodov acknowledges a tension: "doing this with venture, you're still kind of, you know, decapping yourself a little bit... you're just like hobbling on one foot with your hands tied behind your back, right? Like until until you know that the the technology that you are building actually helps you scale into a lot of infrastructure." The operating implication is that Atomic must prove its technology changes the physics of the problem, "if at the core of that technology um you're able to actually change the physics of the problem, those are really interesting opportunities," before the capital structure permits full infrastructure scaling. That constraint shapes hiring: the company needs engineers who can deliver technical proof points fast, not managers who optimize existing processes.
Finance data gets priority; operating data doesn't
A revealing operational maxim comes from McNeill: "Finance data always gets the priority. Operating data doesn't always get that." Atomic's product is built to invert that hierarchy, making operating data (what the machines are doing, what the molds are producing, where the spoilage occurs) as visible and actionable as the P&L. For food-focused customers like DoorDash, that translates to cutting waste and spoilage. The principle is that the factory floor should be as instrumented and decidable as the trading desk.
The values are not posters on a wall. They are the product's architecture: autonomous decision-making, software-native production lines, tribal knowledge captured in model weights, and a business model that only works if the technology genuinely changes the physics of tool-and-die making. A candidate who needs a values document to know how to behave will not last long enough to read one.
What the Hiring Bar Selects For
Atomic Industries screens for two things above all: relevant experience and the ability to operate autonomously. The company's own hiring data puts it plainly — "relevant experience and culture fit is the priority. Show you can work autonomously — that matters more than algorithms at 50-200 people." That priority reflects the reality of a 60-person team where flat decision-making and rapid technical velocity leave no room for hand-holding.
Slodov's own path signals what the bar measures. He didn't read reports — he went to the source. That same obsessive curiosity appears in the company's public messaging: "We're looking for obsessively curious individuals to help us build the future of manufacturing." The interview process is calibrated to detect whether a candidate has that same instinct to go primary, to touch the equipment, to learn by doing.
The team composition tells the same story. That mix, PhD-level simulation talent beside people who have cut steel for decades, only works if every hire can translate across domains without a manager bridging the gap. A recycled CV gets rejected fast; the hiring team notices when an application opens with what the candidate wants instead of what Atomic is dealing with right now and what the candidate would do about it.
Autonomy isn't a buzzword here — it's a survival trait. The company runs on referrals first (internal, then sourced, then recruiter pipelines), with cold applications fourth in line. That means the people who get through the door have already been vetted for self-starting behavior by someone the founders trust. The interview process itself (four stages, roughly 14 days, no take-home assignment) moves faster than the industry median of 18 days for companies this size. Final round is reportedly the most challenging, and it focuses on problem-solving and trade-off thinking, not puzzle questions or algorithmic regurgitation.
The industry context sharpens the filter. With 150-300 applicants in two weeks for typical roles, most applications get ghosted. But the volume also reflects a deeper shift: AI is reshaping what "software engineer" means at companies like Google and Anthropic, where leaders now talk about "builders" who direct AI agents rather than write code line by line. Atomic sits at a different intersection; it needs people who can apply that same agentic mindset to physical production: molds, dies, factory floors, supply chains. The hiring bar selects for engineers who have already operated in that world, or who demonstrate the hands-on velocity to learn it fast.
Slodov put the underlying logic bluntly: "If you don't actually do it in practice, then you forget or you just don't have it... the active pool of people that actually know how to do things is extraordinarily important to bolster, to grow, and to cultivate." The interview process is designed to find people in that active pool, or people willing to join it.
Who Thrives Here, Who Burns Out
The public record on Atomic Industries is thin. Glassdoor lists 23 reviews for the company as of its latest snapshot, a sample small enough that any aggregate rating swings on a handful of opinions. Those reviews are anonymous; the platform surfaces no verbatim excerpts in the research, so the specific praise or criticism employees have written cannot be quoted here. Indeed shows 813 reviews for General Atomics (a separate, much larger defense contractor) and those figures are sometimes conflated in search results. They do not describe Atomic Industries, the Detroit robotics startup.
The Glassdoor aggregate that does exist distills to a single pattern: "smart and driven individuals thrive" in a "dynamic and entrepreneurial atmosphere" that simultaneously offers a "beautiful office" and "collaborative spirit" while demanding "self-direction" amid a "lack of structure." That phrasing, repeated across multiple reviews, is the clearest signal the public record provides about who stays and who leaves.
Slodov's own account of the company's origin reinforces the filter. The company he built reflects that education: it rewards people who have already felt the pain of production, tooling delays, mold revisions, the gap between CAD and steel, and who don't need a manager to translate that reality into daily priorities.
The hiring data on Zero G Talent's board shows three salaried roles posted recently with a salary band of $79k–$212k (median $160k). The spread (General Engineering Applicant at $95k–$225k, Accounting Manager at $130k–$160k, Purchasing & Finance Coordinator at $75k–$85k) suggests a team that pays for outcomes, not titles. The "General Engineering Applicant" listing, open across Detroit, Cleveland, Los Angeles, and remote, is explicitly broad: the company is fishing for builders who can figure out what needs doing and do it, not specialists waiting for a spec.
Slodov has been blunt about the training deficit. He emphasized that there's no time to train hundreds of thousands of people conventionally, noting it would take 20 years, so they must find a way to short-circuit the process. That urgency targets engineers who've already short-circuited their own learning curves: people who taught themselves CNC programming, or ran a mold trial at 2 a.m., or debugged a cooling channel failure without a senior engineer holding their hand. The mission, "re-industrialization" on a 10-to-20-year horizon, also filters out anyone optimizing for a three-year vesting schedule. "There's no shot that you can do something like this in, you know, 2 three years or something like that unless it's war," Slodov noted.
Conversely, the same traits that make Atomic work for some make it untenable for others. Candidates who need onboarding programs, defined career ladders, or a clear separation between "software work" and "shop floor work" will struggle. Slodov dismissed the idea of software-style margins in manufacturing: "Those don't have to have like 90% software margins... that's a complete farce." Engineers who expect clean abstraction layers — push code, watch robots move — will hit the wall of steel, oil, and cycle time. The company's explicit goal to "marry these worlds together" means the people who thrive are the ones willing to stand in both.
The burnout profile isn't a personality flaw; it's a mismatch with the company's current phase. Atomic is pre-scale, pre-playbook, and capital-constrained by venture norms. That means long weeks, context-switching between simulation and the press floor, and decisions made without a safety net. The Glassdoor reviews' "lack of structure" isn't a bug to be fixed — it's the operating condition. People who treat ambiguity as a signal to wait for direction don't last. People who treat it as a vacuum they're paid to fill do.
If you've never had to argue with a toolmaker about draft angles, or explain to a vendor why their quote assumes a process you're trying to eliminate, or stay late because the first article off the mold short-shot and the customer needs parts Tuesday; Atomic will teach you those things, but it won't teach them gently. The ones who stay are the ones who already know the lesson and showed up anyway.
Working in robotics? Zero G Talent tracks the openings: see every open Atomic Industries role, browse robotics jobs, the companies hiring, and the people building the field.