This week the front of the board is quiet and the filings are loud. Let's start where you'd start.
Twenty-seven roles were posted across the Bay Area tech employers tracked here in the last seven days. There are five thousand six hundred and seventy-one open roles in total. Read that the way I do: almost everything you're scrolling past has been sitting there a while, which means the people you're competing with found it weeks ago, not this morning. Fresh postings are where your application actually gets read, and there were twenty-seven of them.
I'm not going to call that a trend, though. There are ten weeks of history here and only five were watched live. The others were reconstructed from job boards after the fact, and a job board only shows you roles still open — so those older weeks under-count postings by design. Comparing this week's twenty-seven to a backfilled week is comparing a photograph to a memory.
Now the layoffs. Two names on the WARN record — that's the California notice a large employer has to file publicly before a big cut, so it's a document, not a rumour. Visa filed for three hundred and twenty roles in San Mateo County, twenty-five days ago. Practically, that puts three hundred and twenty people with payments experience into the Peninsula candidate pool, most of them already authorised to work here. If you're applying to fintech on that side of the bay from abroad, that's who is ahead of you in the queue this month.
Chime cut a hundred and thirty-five, also twenty-five days ago, and has posted sixteen roles since the notice: Product Designer, Product Manager for Lending, Software Engineer on AI Enablement, Software Engineer on Communication Platform, a Design Director. I'm not going to guess why a company does both things in the same month — I don't know, and neither does anyone who tells you they do. What I can tell you is those postings are recent and real, and Chime's open board leans product and design, ten roles there, plus eight in data. Across ninety days: fourteen layoff notices, one thousand three hundred and seventy-nine people. That's the size of the pool you're swimming in locally.
Sponsorship next, because for most of you that's the whole question.
Of the employers tracked here, three hundred and thirty-one have an established record of certified Department of Labor filings — repeated filings, the kind of place where the immigration counsel already has a template and nobody has to be convinced the process is survivable. Six hundred and thirty-six are active. One thousand nine hundred and seventy-four are occasional: a filing here and there, which means ask, don't assume. And a hundred and eighty-one have no sponsorship found.
Slow down for that last one, because it's the easiest thing on this page to misread. No sponsorship found means the Labor Department filings were checked and nothing appeared. It does not mean the company won't sponsor. Plenty of young companies sponsor their first hire with no prior record at all — somebody has to be the first, and that filing exists because a hiring manager wanted one specific person badly enough to pay a lawyer.
Which brings me to the newcomers. Eight employers are new to the tracked set; six are named this week — Lattice, Linear, Elastic, Resend, Runway, Supabase. All six sit in that no-sponsorship-found group. And here's the thing about a newly arrived employer: it's usually competing for far fewer applicants than an established company hiring at the same pace, and that advantage decays within weeks. So the play is to apply now, and put the visa question in the first recruiter conversation, plainly, in one sentence. Not in the cover letter. On the call, where a human can say "let me check."
Now, OpenAI. Their hiring pace is accelerating, and it's the largest move anywhere in this week's data — nothing else in the file comes close. It also happens to be the biggest board here. Twenty-nine open roles in AI and machine learning. Twenty-seven in infrastructure and platform. Twenty in product and design, fifteen in security and trust, fourteen in data, nine in frontend and mobile, eight in backend and APIs, seven in engineering leadership, six in forward-deployed and solutions. They're at or near the top of almost every area measured. If you've been treating OpenAI as one application, you're underusing them — that's at least eight different hiring managers with different problems.
The counterweight: Anthropic has slowed sharply over the same four weeks, and so has Brex. A slowdown usually means a hiring push is finishing, not that a company is in trouble, and the open boards say the same. Anthropic still has twenty-nine infrastructure roles open — the most of any single company in that area — plus twenty-three in AI and machine learning, eighteen in security, twelve in engineering leadership. Brex still has sixteen product and design roles, second only to OpenAI. Slowing down from a sprint is not the same as stopping.
Then there's the money that hasn't turned into jobs yet, and there's a lot of it.
Whatnot disclosed selling five hundred and forty-seven million dollars under an SEC Form D seven days ago and has posted nothing since. Baseten disclosed just over one point zero nine billion, fifty-six days ago — nothing since. Core Automation, four hundred and thirty-two million, twenty-six days. CodeRabbit, a hundred and forty-three million, eleven days. Generalist AI shows up twice: three hundred and sixty-four million eighty-one days ago, and another hundred and ninety-eight million one day ago. Still no postings. To be precise about what these are: that's the total reported sold under an offering, not necessarily one round, and the filings don't name a round letter, so I won't either.
Here's why I'd reorganise a Monday around this list. The typical gap between a filing like that and the first job posting is thirty to ninety days. Whatnot is on day seven. Generalist AI is on day one. So if you find an engineer at one of those companies on a public profile this week and send four sentences about what you'd build — you are not competing with four hundred other applicants, because the posting that generates those four hundred applicants does not exist yet. Mercury is the exception to keep honest: two hundred million disclosed, ninety-seven days ago, nothing posted. That's past the usual window, so treat it as a watch item rather than a warm door.
One last thing, and it's the number I've been putting off. Across roughly thirteen hundred open roles that got a skills read, three hundred and twenty-four are explicitly senior, two hundred and sixty-one are staff or above, two hundred and fifty-four are management — and twenty-six are entry level or new grad. Twenty-six. I checked it twice. If you're two years into your career, this board is mostly not addressed to you right now, and I'd rather say that than dress it up. Your two openings are the newcomers I named and the newly funded companies above, because at both, nobody has finished writing down what the headcount is supposed to look like.
If you want depth rather than breadth, the specialist employers are worth a note. Sierra AI is accelerating, and twenty of their fifty-one open roles are in AI and machine learning — nearly two in five, so you'd be doing that work, not adjacent to it. Scale AI, fifteen of fifty-six. Notion Labs, ten of fifty-three in infrastructure. Cloudflare, six of sixty in forward-deployed and solutions, which is the engineer who sits with the customer and makes the thing actually work in their environment. One caveat that matters: all of that comes from job titles, not descriptions — descriptions aren't stored here — so treat it as a strong hint about a team's shape, then confirm in the interview.
So Monday: the six newcomers, and the Form D list. Those are the two places this week where showing up early still buys you something.
One quick thing before the fine print. Everything behind this episode — the San Francisco Bay Area board, the companies I named and the ones I ran out of time for — is at hiringclimate.com. It is free, and it is the same data I read to make this. And if this was useful, like the show or follow it wherever you are listening. That is genuinely how other people find it.
One last thing, and it is about this show rather than about the market. This episode was written by an AI system, and the voice reading it to you is synthetic. The facts underneath it are not: every number here comes from a public filing, a job board or a government record, and it can be checked. But nobody sat down and wrote this, and you should know that before you act on it.