The California hiring picture is already split in two

Recruiters in the Bay Area are working two opposite markets at the same time.

US tech job listings sit roughly 36% below their February 2020 baseline. Inside that number, machine learning engineer openings are up 59% while general software engineering positions are down 49%. Android, Java, iOS, .NET and web developer roles each fell more than 60%.

At the same time, tech led May 2026 hiring announcements with 11,250 planned new positions while remaining one of the year's largest sources of job cuts. Resume.org found 92% of hiring managers expect to hire in 2026 and 56% expect layoffs at their own company. The roles leaving and the roles arriving are not the same roles.

For a California employer, that means your req is competing against a small pool that has options, while your inbox fills with applicants for the roles you are no longer opening.

Five roles worth funding after Astra

AI systems engineer. Owns the agent stack: tool definitions, evals, retries, cost ceilings and fallbacks. This is the role that turns a model demo into something that survives production traffic.

Applied AI product manager. 47% of AI product postings are manager level, so the entry route is narrow and the competition is senior. This person decides which workflow gets automated and which stays human.

Security engineer with AI exposure. Given Astra's cyber capability rating and the default-off enterprise posture, security review is now part of every AI rollout rather than a later stage.

Data engineer. Model quality tracks data quality. Retrieval, permissions and pipeline hygiene decide whether the agent is useful or confidently wrong.

Senior reviewer, not junior producer. If a model drafts the first version, the scarce skill is the person who can tell a good version from a plausible one. Hire for judgment and taste at the review layer.

What to change in your hiring process

Rewrite the take-home. If a candidate can finish your assignment with Astra in fifteen minutes, the assignment tests nothing. Replace it with a review exercise: hand them AI-generated code or a strategy doc with three real flaws inside and ask what they would ship, what they would cut and why.

Interview for cost awareness. At $50 per million output tokens, an engineer who leaves an agent looping is an expensive engineer. Ask candidates how they would cap spend on an agentic workflow.

Stop screening on tool familiarity. Astra shipped in September 2026. Nobody has two years of experience with it. Screen for fundamentals and the speed at which someone picks up a new stack.

Shorten the loop. 65% of tech hiring managers report that finding skilled talent is harder than a year ago. Senior AI candidates in California usually hold more than one offer. A five-stage process spread over six weeks loses them to a three-stage process that closes in ten days.

Consider training up. 57% of technology hiring managers say they prioritize high-potential candidates and develop them when the exact skills are not available. With AI skills changing faster than most people can learn them, this often beats waiting for a perfect profile.

Where a specialist hiring agency earns its fee

93% of tech and IT leaders told Robert Half that staffing firms were effective at helping with AI-related hiring challenges. The reason is access and speed rather than volume of resumes.

A California hiring agency worth using should give you three things. First, a live compensation read from actual placements in your metro, because a San Francisco range and a San Diego range are not the same conversation. Second, passive candidates, since the strongest AI engineers in the state are employed and not applying. Third, an honest read on whether your req is realistic before you spend eight weeks proving it is not.

Ask any agency you are evaluating for their average time to fill on AI and ML roles in California, and the offer acceptance rate that goes with it. Fill speed without acceptance rate tells you nothing.

Frequently asked questions

Will GPT-6 Astra reduce tech hiring in California?

Not in aggregate so far. AI vacancies in the US rose 36.0% year on year to 55,374 in Q1 2026, and California holds the largest share. What is falling is demand for general software roles, which dropped 49% against the 2020 baseline. Budget is moving between categories rather than leaving.

What salary should a California employer budget for AI roles in 2026?

The national median for AI roles reached $162,240 in Q1 2026, 22.4% above non-AI IT roles. California, and San Francisco in particular, prices above that median. Use placement data from your own metro rather than a national average.

Should we hire for GPT-6 Astra experience specifically?

No. The model launched on September 3, 2026, so any claimed depth of experience is thin by definition. Hire for agent design, evaluation and system thinking, which carry across model releases.

How long does it take to fill a senior AI role in California?

Plan for a competitive close rather than a long search. Candidates typically hold multiple offers, so interview stages and offer turnaround decide the outcome more than sourcing volume does.

Next step

If you are planning headcount for the rest of 2026, start by sorting your open reqs into two lists: roles where a model can now do the first draft, and roles where someone has to decide whether that draft ships. Fund the second list first.

Talk to our team for a compensation benchmark and a shortlist for your California AI roles.