Engineers placed close to a customer’s work, known as forward deployed engineers or FDEs, are in demand. Postings for the job rose 729% in a year, to 5,330 in April 2026, according to Indeed data reported by Business Insider Taiwan. The pay shows the same pressure, with OpenAI listing a San Francisco FDE role at $162,000 to $280,000 plus equity.
The title sounds like a military metaphor, which makes the job easy to misread. The engineer studies how the customer’s organisation runs, writes code inside its environment and connects the work to existing systems.
Palantir, the data software company, defined the role. A 2026 essay from the venture firm a16z says Palantir gave its solutions engineers and integration engineers the forward-deployed title in 2011. The branding worked because it raised the status of work that had usually sat below central product engineering.
AI has made that old pattern newly valuable. Many companies can buy access to capable AI models. The model alone does not know the business process, the trusted data source, the approval chain or the informal workarounds that keep an organisation running. Customer teams need someone who can turn those realities into working software.
That person sits between two familiar roles. A traditional software engineer can build but often lacks daily access to the customer’s messy operating reality. A consultant can lay out the problem but may leave the client with recommendations while another team owns the build. The FDE role works when one person can diagnose the workflow, earn trust, ship the first useful version and keep improving it after users react.
The hard part is often organisational. The engineer has to find where the trusted data lives, which team owns it, who will use the tool and which manager can unblock access. That takes technical judgement to build well and social judgement to make the system acceptable to the people who will depend on it.
Job listings describe the same mix. Salesforce says its Product FDE team works alongside major customers to test new AI capabilities and turn what it learns into repeatable patterns. OpenAI’s listing has FDEs owning the whole job with major customers, from working out what to build to designing, building and rolling it out. It says success is measured by real use, the effect on customers’ work and feedback that changes OpenAI’s product and model plans.
