The Surge Isn't Quiet
HappyRobot lists 81 open roles on Figured, a board that mirrors its applicant tracking system in real time; 31 posted in the last 30 days alone per Figured, with salary bands stretching from $65,000 to $500,000 and a median of $225,000 per Zero G Talent's board. Zero G Talent's board carries 30 live listings overlapping the same roles. Two landed in the past week: a Strategic Account Executive at $400,000–$500,000 and an Enterprise GTM Lead at $375,000–$400,000, both based in San Francisco.
The listings refresh hourly. The volume has stayed elevated for weeks across eight offices on three continents: San Francisco, Chicago, New York, Mexico City, Madrid, Barcelona, London, and Sydney. Every opening groups under five departments: Deployment, Engineering, Go-to-Market, Operations, and Revenue. That spread signals a build-out across the full stack, not a concentrated push in one function. LinkedIn amplifies the signal: 35,400 followers on the company page, and each new requisition surfaces in the feeds of candidates tracking AI agent startups. The pace makes clear the company is not backfilling. It is scaling, and its screening process, emphasizing specific AI engineering skills and cross-functional collaboration, is already reshaping how candidates tailor resumes.
What the Listings Actually Screen For
HappyRobot's open roles cluster around two tracks — engineering and go-to-market — but the screening criteria converge on a single profile: engineers who ship product end-to-end and GTM operators who think like builders.
For software engineers, the baseline is full-stack fluency with a declared preference for TypeScript and Next.js on the front end and Go on the back end. The listings don't ask for familiarity. They ask for "front-end expertise with TypeScript and Next.js, including performance, UX, and clean architecture" and "back-end experience with Go, building reliable, scalable systems." Candidates must "design, implement, and ship features end-to-end, from UI to APIs to infrastructure" and "help optimize and scale our voice AI infrastructure, improving latency, reliability, and cost." That last line is the differentiator. HappyRobot built its own voice stack, models, and orchestration layer from the ground up, starting in logistics.
The "strong generalist" framing appears repeatedly ("excellent on either the front end or back end, and comfortable working across the stack"), but the ownership language pushes past generalist into product engineer territory. Listings require candidates to "build and own entire sections of the product, including major parts of the web application," "collaborate closely with founders and teammates to shape product direction," and "ship to production frequently and iterate based on real user feedback." The phrase "product-minded engineer who understands tradeoffs between speed, quality, and scale" reads as a filter against resume-driven development. The company wants evidence of shipping judgment, not framework checklists.
On the GTM side, the newest roles — Strategic Account Executive ($400k–$500k, Zero G Talent's data shows), Enterprise GTM Lead ($375k–$400k, Zero G Talent found), and two GTM Recruiter positions ($180k–$250k each, according to Zero G Talent's board) — signal a shift from founder-led sales to a structured pod model. The Enterprise GTM Lead listing spells out the screen: "This role combines hands-on execution with leadership, working cross-functionally with Product, Engineering, Marketing, and Ops to scale repeatable GTM motion" and "Own the go-to-market strategy and execution for your pod, translating company-level objectives into clear plans that drive revenue, adoption, and retention." That cross-functional language mirrors the engineering listings almost word-for-word, suggesting the same cultural filter applies regardless of function.
The cultural values listed on the careers page operate as explicit screening rubrics. "Extreme Ownership," "Craftsmanship," "Urgency with Focus," "Talent Density and Meritocracy," and "First-Principles Thinking" appear as bullet points under "How we work," not as aspirational fluff. The "First-Principles Thinking" section reinforces the signal: "We don't copy-paste solutions. We go back to basics, ask why things are the way they are, and rebuild from the ground up if needed."
The net effect is a screen that selects for builders who have operated at the infrastructure layer of voice AI or adjacent real-time systems, and for GTM operators who have sold technical products to enterprise buyers while sitting beside engineering teams. The listings don't ask for years of experience. They ask for scope of ownership and evidence of shipping in environments where latency, reliability, and cost are first-class constraints.
The Five-Second Filter
The volume of inbound applications has not translated into a proportional pool of qualified candidates. In a widely viewed recruiter breakdown, Brian from A Life After Layoff estimated that roughly one in twenty applications are "actually qualified and would get a recruiter excited to speak with them." The rest, in his words, are "really kind of total garbage." That figure, consistent with what hiring managers describe across the AI talent market, frames the screening problem: candidates have about five seconds to land in the yes pile before a recruiter dispositions them.
Glassdoor data from HappyRobot interview candidates shows 16 posted interview questions and 20 reviews, a modest but specific signal that the process is structured enough to generate a traceable footprint. One Content Manager candidate reported being asked, "what my job search experience has been like," an unusual opener that tests narrative control and self-awareness rather than technical depth. The same recruiter breakdown noted they know within the first ten minutes whether a candidate fits the role. The resume, they emphasized, is "your best and probably your only first impression."
Interview performance carries disproportionate weight. The recruiter breakdown noted that "the best person doesn't always get hired for the role"; candidates with "the Gift of Gab" who "check all the boxes" can convince a hiring team they're the better fit even when someone else looks stronger on paper. Recruiters advocate for those candidates: "a lot of times we will like you as a candidate and we'll think that you check all the boxes and we will go and advocate for you to the hiring manager." But the hiring manager retains veto power and may eliminate a candidate after reviewing interview notes before a conversation ever happens.
Speed signals intent. Recruiters are graded on time-to-fill, so long delays after interviews are "generally not a great sign"; it means they're interviewing others or undecided. Conversely, "home run" candidates have received offers the same day as their first round. When a hiring manager doesn't want to lose a standout to a competing offer, "they're gonna make an offer happen very quickly." Salary bands stretch for those candidates: "I've seen candidates come in with salary expectations way above the range that we could afford but they are such strong candidates that the hiring manager went to bat for them and figured out a way to get it done." Tactics include combining multiple roles into one headcount and increasing equity stakes.
Candidates who understand the recruiter's incentives — speed, fill rates, hiring-manager advocacy — and the hiring manager's veto power are the ones who convert. They don't just match the spec. They make the case easy to champion.
Why the Market Is Breaking
The screening intensity at HappyRobot reflects a market where compensation for AI specialists has detached from traditional software engineering bands and entered territory that would have seemed implausible three years ago.
| Role / Tier | Reported Range | Source / Context |
|---|---|---|
| Senior AI research scientist (large tech) | $500,000 – $2,000,000 | Harrison Clarke, via FT; up from $400K–$900K in 2022 |
| Senior AI scientist (top tier) | $3,000,000 – $7,000,000+ | FT, citing industry recruiters and job-movement data |
| AI engineer at Meta | $186,000 – $3,200,000 | Levels.fyi data |
| AI engineer at OpenAI | $212,000 – $2,500,000 | Levels.fyi data; median higher than Meta |
| Machine learning engineer (US average) | $175,000 | Indeed, 2025 |
| Machine learning engineer (US high end) | ~$300,000 | Indeed, 2025 |
| Machine learning / principal engineer (London) | £140,000 – £300,000 | Robert Walters, via CNBC |
| Senior software engineer (non-AI) | $180,000 – $220,000 | FT / Economic Times |
| HappyRobot roles (board median) | $225,000 | Zero G Talent's board; band $65K–$500K across 30 listings |
The spread is deliberate. Meta's push — accelerated after Llama 4 underperformed on reasoning and coding benchmarks — has forced every other lab to recalibrate. Mark Zuckerberg reportedly offered $100 million signing bonuses to OpenAI staff, and a $250 million package to a 24-year-old researcher, Matt Deitke, who left a University of Washington PhD program. Zuckerberg later told The Information that "the amount that is being spent to recruit the people is actually still quite small compared to the overall investment… when you talk about super intelligence." The logic is cold: if a frontier model costs $1 billion to train, a $10 million engineer is a rounding error.
"It has just become manically more hyper intense over the past few years, to the point where it feels like certain players are willing to do anything or whatever it takes to bring that talent into the organisation," said Kyle Langworthy, partner at Riviera Partners, speaking to the FT.
The ripple hits three concentric circles. First, startups. OpenAI's chief research officer Mark Chen described recent departures as "someone has broken into our home and stolen something." Riviera's Langworthy added: "It can be extremely challenging to hire your AI, engineering, and product team when you're a lesser-known company." Acqui-hiring has become a primary talent channel: large companies buying startups not for product but for the engineering team. Golden handcuffs (four-year vesting schedules) lock in the few who stay.
Second, academia. Universities are being emptied of professors; corporate packages dwarf academic salaries by an order of magnitude. The pipeline of PhDs from the top five or six global programs is being snapped up before graduation, per Ben Litvinoff at Robert Walters. "Experience required to train models with trillions of parameters is only gained in the field," noted The AI Chronicle; meaning the university system can no longer produce ready-to-deploy talent at the frontier.
Third, geography. Europe faces a severe brain drain. Thomas Wolf, co-founder of Hugging Face, put it bluntly: "If you take one software engineer in the Bay Area right now, you can have three to four people of roughly the same level in Europe." Hugging Face is now recruiting heavily in Europe; German startup Aleph Alpha expanded sixfold in a year by pitching research freedom, publication rights, and impact-driven work. But the compensation gap remains structural: US giants pay Silicon Valley rates remotely, hollowing out local ecosystems.
Traditional industries are collateral damage. Mark Miller, CEO of Insurevision.ai, told Startups Magazine the talent war creates a "massive opportunity gap" in insurance, healthcare, and logistics. "Entire industries… can't compete on salary. They need innovation but can't access the talent. The current situation is absolutely unsustainable."
Competitors are responding. OpenAI told staff it is "recalibrating comp" and scoping "creative ways to recognise and reward top talent." Meta invested $15 billion in Scale AI and brought Alexandr Wang in to lead a "superintelligence" team. Some firms are partnering with universities, launching specialized AI degrees, or lobbying for government-funded research centers to retain domestic talent. The EU's AI Act proceeds while the talent exodus accelerates: a regulatory framework chasing a workforce that has already moved.
The market has bifurcated. At the top, a few labs bid up a slim talent pool of researchers who have actually trained trillion-parameter models. Everyone else (HappyRobot included) competes for the next tier, where median pay sits around $225,000 but the ceiling is rising fast. The moat is widening.
The Kicker
For the 81 roles still open on Figured, the five-second filter keeps running. The next resume that lands in the yes pile will have front-loaded Go and TypeScript, quantified latency gains, and a story about throwing out an architecture, because at HappyRobot, the screen isn't looking for AI enthusiasts. It's looking for builders who have already shipped at the infrastructure layer.
Working in frontier tech? Zero G Talent tracks the openings: see every open HappyRobot role, browse frontier tech jobs, the companies hiring, and the people building the field.