There are two doors into a frontier AI lab.Only one of them is open to you.
Anthropic and OpenAI are hiring on two completely different tracks right now. Knowing which one you're actually near could save you months chasing the wrong door.
Why nowOne program converts 40%+ of fellows to full-time
If you read nothing else
Anthropic and OpenAI are hiring on two tracks at once: poaching already-famous people at the top, while running structured programs — Fellows, Residency, Claude Corps — that convert 40%+ of participants to full-time staff at the bottom. The second track is the one you can actually walk through. It still requires proof, just not pedigree.
🎙️ Nova's Signal
Nova is FOWL AI's news anchor — she tracks the signals every week.
"Every 'how I got hired at OpenAI' story you've read skips the boring version: most people who work there were already impressive somewhere else first. But two specific doors don't require that — and almost nobody talks about them, because they don't make for a viral post."
Nova's call: This is the most actionable issue we've run in weeks. Save the deadlines section.
Every "day in the life at a frontier lab" post and "how I got hired at Anthropic" thread makes it sound like there's one path in: be brilliant, build something, get discovered. That's true for a small number of people — and it's not the whole story.
Underneath the viral hiring stories, there are two very different hiring tracks running at once inside these labs. One is basically closed unless you're already somebody. The other has a published conversion rate, explicitly welcomes people with zero traditional AI background, and is open for applications right now.
Here's what we found going through named 2026 hires, the labs' own program pages, and a breakdown of 1,680 real engineer résumés.
5 Doors In, Ranked by How Open They Actually Are
01
The Résumé Door — and it's basically shut
One sentence: Anthropic's highest-profile 2026 hires were all somebody else's star player first — poached at the top, not grown from the bottom.
The named hires read like a most-wanted list: Andrej Karpathy (OpenAI co-founder, ex-Tesla Autopilot lead) to build a new pre-training team; Eric Boyd (17 years at Microsoft, president of AI Platform) into infrastructure; Tom Blomfield (co-founder of Monzo and GoCardless) onto the compute team; Jelani Nelson (chair of UC Berkeley's CS division); Chris Ciauri (ex-Google Cloud and Salesforce exec) to scale Europe and Asia; Sarah Heck (Stripe and the White House National Security Council) to lead policy; Ross Nordeen (an xAI founding member) into security.
Why it matters · This is the hiring track that makes headlines — and every name on it already had a title most people spend a career trying to reach.
The signal — If a hiring story starts with "[Person] joins from [famous company] as [VP/Chief title]," you're reading about the closed door, not the open one.
02
The Door Nobody's Writing About: Enterprise Sales
One sentence: OpenAI and Anthropic have pulled roughly 100 people from Salesforce alone in 2026 — most of them into sales and go-to-market roles, not research.
Reports put it at 45 hires to Anthropic and 40 to OpenAI over the last six months, primarily in sales, marketing, and go-to-market functions. The reasoning is pragmatic, not technical: these labs are selling into the same Fortune 500 accounts Salesforce has spent two decades building relationships with, and that experience is worth more to a lab's growth right now than another research paper. It lines up with what's already visible on FOWL AI's own jobs board — roles like Trust & Safety Specialist and AI Policy & Governance Analyst don't ask for a machine learning background either.
Why it matters · If you're a mid-career enterprise sales or customer-success professional, you may be closer to a frontier-lab job than most "break into AI" content will ever tell you.
The signal — None of these roles required an ML background. They required an existing book of enterprise relationships.
03
The Door That Actually Converts: Fellowships and Residencies
One sentence: two structured programs publish real numbers on this, and both explicitly welcome people with no traditional AI background.
The Anthropic Fellows Program runs 4 months, pays a $3,850/week stipend plus roughly $15,000/month in compute funding, and — in Anthropic's own words — "over 40% of fellows subsequently joined full-time." The OpenAI Residency runs 6 months at $18,333/month as a full employee, not an intern, and OpenAI's own FAQ states plainly: "For those Residents who perform well we extend full-time offers to join OpenAI immediately upon completion." Past residents include someone who came from biology with zero AI background. A third program, Anthropic's Claude Corps, pays $85,000/year and explicitly requires no college degree or prior experience.
Why it matters · This is the one door with a published, real conversion rate. It's not a maybe.
The signal — The Fellows Program's November 2026 cohort closed applications on July 26; later cohorts are accepted on a rolling basis after that, so this is worth tracking, not a one-time miss.
04
The Proof-of-Work Door — real, but rarer than it sounds
One sentence: Gabriel Petersson dropped out of high school at 17 in a Swedish town of 5,000 people and is now a researcher at OpenAI — but the honest version of his story is less "anyone can do this" than it looks.
Petersson got noticed by building things before anyone asked him to: a week of 16-hour days building a working demo site for Midjourney, unpaid, before they'd even talk to him, which got him hired there in 2023. OpenAI actually rejected him once before a friend's referral got him in months later, in December 2024. He's 22 now, earning six figures. It's a real story — but a breakdown of 1,680 Anthropic engineer résumés found that even among their junior hires (under 6 years' experience), 50% had already interned at Meta, Google, DeepMind, Microsoft, or Amazon. "No degree" doors are real. "No proof" doors aren't.
Why it matters · This door exists — but it asks for a portfolio or a founder story, not the absence of one.
The signal — Petersson's own advice: "Anyone can compete if you just put yourself in the right scenarios." True, but read alongside the résumé data above, not instead of it.
05
The Door Being Built, Slowly
One sentence: Anthropic is investing in a wider pipeline for later — it's just not a hiring channel yet.
Anthropic partnered with CodePath, the largest provider of collegiate computer science education in the US, to give more than 20,000 students at community colleges, state schools, and HBCUs access to frontier AI tools and industry-vetted courses — over 40% of those students come from households earning under $50,000 a year. Combined with Claude Corps' no-degree-required $85K track, it reads as a deliberate attempt to widen who even has a shot at the doors above.
Why it matters · Neither program is a current mass hiring channel — but both are evidence the labs know their existing pipeline is too narrow, which is worth watching over the next year.
The signal — Access to tools and courses isn't the same as a job offer. Treat this as a door being built, not one you can walk through today.
📡 Signal & Chatter
What the data and community are saying this week.
›
Anthropic Fellows' November cohort closed today. If you missed it, later cohorts are rolling — worth setting a reminder rather than writing the program off.
›
OpenAI Residency pays $18,333/month as a full employee, not an intern stipend. Worth knowing before you assume these programs don't pay real money.
›
Perplexity nearly tripled headcount in a year — roughly 500 to 1,400+ — and still says it favors applied engineers who ship over pure research pedigree, with 60 roles open right now.
›
"Entry level" at a frontier lab rarely means actual entry level. Half of Anthropic's junior hires already had a FAANG internship on their résumé before they applied.
🔮 FOWL Prediction #10
"By Issue 20 (early 2027), at least one frontier lab will publicly report its Fellows- or Residency-style conversion rate as a recruiting metric — the same way startups report application-to-funding stats."
Anthropic already volunteered its 40%+ number without being asked. Once one lab treats a strong conversion rate as a selling point, the others will have a reason to publish theirs too. We'll check back on this prediction when Issue 20 goes out.
FOWL AI · July 27, 2026 · We'll score this in early 2027.
✅ 3 Things to Do This Week
One for each type of reader — pick yours
01
If you have zero traditional AI background — look specifically at Fellows/Residency-style program pages, not general job boards. These are the doors that publish real conversion numbers and explicitly welcome your background.
02
If you're in enterprise sales or customer success — don't rule yourself out because your résumé isn't technical. ~100 people just like you moved from Salesforce into these labs this year alone.
03
If you're planning to "build something to prove it" — it's a real door, but the honest odds say pair the portfolio with some form of credential-building, not instead of it — half of even junior hires already had one.
🚀 From the Lab
Sponsored by FOWL AI
Launch Lab — Building something?
We're not an agency — we don't build it for you. Brand Brain is the AI workspace you use to turn what you know into a website, newsletter, and digital products you actually own. You stay in control of it the whole way.
Both OpenAI and Anthropic have this title open right now, and it sits squarely in the "door nobody's writing about" category from this issue: it doesn't ask for a research background, it asks for judgment. The role involves reviewing how a model's outputs affect real users, deciding where a product needs guardrails, and working across policy, product, and engineering to get them built.
People moving into this role come from content moderation, trust and safety at other tech companies, law, and policy — not machine learning. It's one of the clearest examples in this issue of a real, current opening at a frontier lab that doesn't require you to already be famous or hold a PhD.
How to position for this now: If you've done policy, compliance, content moderation, or trust and safety work anywhere — even outside tech — lead with that directly. It's more relevant here than a general "interested in AI" pitch.
The honest version of "how do I get hired at a frontier lab" isn't one answer. It's two tracks running at once — one closed to most people, one genuinely open to people willing to walk through it with proof in hand.
Next time you read a hiring story from one of these labs, ask which door it's actually describing before you decide whether it applies to you.
"The path in isn't closed. It's just not the path everyone's describing."
💬 One question — reply and tell us
Which door are you actually closer to right now — the fellowship, the sales pipeline, or building something to prove it? Hit reply and tell us — we're tracking this for a follow-up issue.
Hit reply. We read every one and it shapes what we cover next.