One of the fastest-growing jobs in Silicon Valley sends engineers straight to customers’ offices to get AI up and running—and pays more than $188,000
Fortune Preston Fore
Silicon Valley is hiring engineers to work on AI inside customers’ offices, not just from HQ. The job is exploding, and pay can top $188,000, because getting AI to actually work is the hard part.
Based on reporting by Fortune, Preston Fore — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Silicon Valley’s newest hot job is less about building AI in a sleek office and more about hauling it into the mess of real businesses. Forward-deployed engineers go onsite, wire up tools, connect software, and try to make AI fit actual workflows. It’s a role that looks part engineer, part translator, part fixer.
The numbers are wild. Lightcast found postings for forward-deployed engineers rose more than 1,000% from January to August 2026 versus the same stretch a year earlier, and more than 4,600% versus 2023. That is way ahead of the broader tech market, where job postings were up 13% year over year. Similar growth has also shown up on LinkedIn and Indeed, according to The New York Times.
Palantir helped make this style famous. The company has around four dozen open forward-deployed roles right now, mostly in software development, and says candidates can end up working with clients such as Intel, NATO, and Norway. Its job listing describes the work as something close to a startup CTO’s life: small teams, little supervision, and end-to-end ownership of high-stakes projects.
The pay helps explain why people are paying attention. Lightcast data shared by Dice president Paul Farnsworth put the median advertised salary for these roles at more than $188,000, compared with roughly $145,000 for traditional software engineers. At some firms, including Anthropic, compensation for certain forward-deployed jobs can reach $400,000. And the model is spreading well beyond Palantir, with Microsoft, Meta, Google, OpenAI, and Anthropic all posting for these roles.
What employers actually want is not just AI fluency, but the ability to make AI useful. Farnsworth said the key is combining technical basics like APIs, data pipelines, cloud infrastructure, and getting systems into production with consulting chops, communication, and business judgment. The blunt version: knowing the latest model is nice, but making it do something useful inside a company is what gets hired.
My take — AI-written commentary, not fact-checked reporting
This is the sort of job boom that says more about corporate chaos than AI magic. If a model needs a human courier to survive contact with a company’s own software, the product is still half-finished. Silicon Valley can call it “forward-deployed” all it wants; most of the time it just means the integration bill arrived in a blazer.
Read more about this at: Fortune