Why This Field Matters

Forward deployed engineers used to be hired one at a time. One person who sits inside a customer and gets the AI into real work, then another. The layer above that is now opening. Vercel’s careers page lists a Director of Forward Deployed Engineering next to its individual FDE openings, and its solutions architect ladder runs individual, manager, director. Replit posted a Head of Forward Deployed Engineering on its public job board at 12 or more years of experience and a 300,000 to 400,000 dollar band. OpenAI’s careers page carries a role with a long name: Manager, Technical Deployment Leads (TDL), Forward Deployed Engineering (FDE), asking 8 or more years.

A title moving up a rung means the work has become repeatable. When a deployment is a heroic one off, you do not need a leader, you need your best person on it. Once the same pattern runs twenty times, different questions show up. Who gets staffed onto which account, and for how many weeks. Does the customization this account asked for become a product feature, or does it stay in that account forever. Of everything sales promised in the contract, how much can this team actually absorb.

Cursor and Databricks are hiring FDEs and solutions architects side by side. The names vary by company, but the org charts converge on the same shape: a deployment organization sitting between the people who sell and the people who build, carrying its own P&L and its own headcount.

Required Skills

Leading here does not mean leaving the code behind. The reason these postings still ask for 8 to 12 years is that only someone who can read a pull request can hold this team’s judgment calls. Everything else sits on top of that.

• Account portfolio management. The cost base of an FDE org is people time. Who you staffed where, for how many weeks, is your margin. A team that cannot see hours per account has no idea which customer is losing money.
• Deciding what becomes product. When a third customer asks for the same thing, that is not a customization, it is a feature. Negotiating that boundary with the product org, and then moving code ownership to match, is the core of the lead’s job.
• Hiring and an internal pipeline. People who write production code and also sit in front of customers are thin on the market. Fill only from outside and roles stay open for months, so you design the path that moves product engineers into deployment work yourself.
• Scoping before the contract. A pilot’s scope can only be cut before it is sold. If nobody in the sales call says “that is six months, not six weeks,” the entire gap lands on the engineers.
• A standard path through security and compliance. On premise installs, SSO, audit logging, data export review. Do it from scratch per customer and throughput stays flat no matter how many people you add.

Career Path

The entry point is senior FDE. You need to have owned two or three accounts end to end, and taken at least one of them past the pilot into rollout, before the next rung opens. In the lead band you run three to five people against a single industry vertical, and what you prove there is not personal output but a record of other people’s deployments staying up. At the head or director level, org economics and the boundary with the product team take most of the week.

In the US market the pull comes from two directions: AI vendors institutionalizing what Palantir invented, and go to market heavy startups that have to win their first large accounts by hand. Compensation reflects the scarcity, as the Replit band shows, but so does the risk. Ask how the deployment org is measured. If it is measured on bookings alone, the engineering side of the job quietly turns into presales.

If you only get to ask one question in the interview, ask this: of everything the team built for customers over the past year, how much of it landed in the product. If the answer is close to nothing, that org is internal consulting, and the leadership seat is a consulting manager seat.