Forward deployed engineer job description

A job description for this role has one job: attract the people who will thrive and repel the people who will resign in month four. Most of them do the opposite by being vague about the two things candidates care most about.

The two paragraphs that matter

Which job is this? The analysis of 1,000 postings found the title covers three different jobs: a production engineer embedded with customers at 60 percent, a sales engineer with implementation duties at 30 percent, and an internal tools builder facing no external customer at 10 percent. Candidates know this now. A description that does not say which one it is reads as either careless or evasive.

Say it directly. "You will spend roughly 80 percent of your week writing production code in customer repositories" is a sentence that sorts your funnel for you.

How much travel? 68 percent of postings required travel, and where a figure was given, 50 to 75 percent was typical. OpenAI states up to 50 percent on its own postings for the role.

Hiding this does not increase acceptances, it increases early attrition. Put the number in the first hundred words.

A template

Adapt it. The comments in brackets are instructions rather than copy.

Title: Forward Deployed Engineer

One-line summary. You will embed with one or two of our enterprise customers, build production software inside their systems, and stay accountable until it runs. [If this is not true, use a different title.]

The shape of the week. Roughly 80 percent building, 20 percent with the customer. [State your real number. If it is 40/60, say so and call the role something else.]

Travel: [Your number] percent, in blocks of [days or weeks] at customer sites in [regions].

Compensation: base plus equity, no sales quota. [Across 1,000 postings, 70 percent mentioned equity, 8 percent mentioned commission, and none were quota-carrying. If yours carries a number, say so here.]

What you will do

• Work inside a customer's environment, on their real data and their real approval process
• Write and ship production code, reviewed and merged in their repository or ours
• Run discovery with the people who will use what you build, and write down what correct means
• Take a system from nothing to running in production, including the security review and release path
• Identify what generalises and feed it back to product [name the cadence]
• Hand the system over to the customer's own engineers on an agreed date

What we need

• A real engineering foundation. You will clear the same technical bar as our product engineers
• Comfort working without a specification, and the judgment to write one
• The ability to scope a vague problem out loud, in front of people who disagree with each other
• Willingness to say no to work that only ever serves one customer
• Genuine curiosity about how a business actually operates

What we do not need

• Deep expertise in one narrow stack. You will be in unfamiliar systems constantly
• A background in consulting delivery, though it is not a barrier
• [Be honest here. This section builds more trust than the requirements list.]

Stack: [List it. Python appeared in 66 percent of postings and TypeScript in 35 percent, so if yours is unusual, say so early.]

What is hard about this job. [Include this section. Name two real difficulties. Candidates trust a description that admits them, and the ones who are put off were going to leave anyway.]

What to leave out

"Rockstar", "ninja", "wear many hats". The last one is code for undefined scope, which is the thing this role already has too much of.

A ten-item requirements list of specific technologies. The role needs generalists who ramp in unfamiliar systems. A long stack list selects for the opposite.

"Fast-paced environment" as a substitute for the travel number. Candidates read this as an admission you are hiding something.

An inflated seniority label. The data says this role is staffed mid-level: 60 percent of postings targeted three to five years, 20 percent six to eight. Calling it Staff to attract applicants creates a levelling problem you will inherit.

The experience level to actually target

From the same dataset: 12 percent of postings targeted zero to two years, 60 percent three to five, 20 percent six to eight, 8 percent nine or more.

Alongside that, First Round Review reports that ten or more years at a large tech company can be a red flag for this work, because the role needs people who operate without institutional scaffolding. That is a statement about a specific background rather than about seniority itself.

Routes in, for sourcing: 45 percent from software engineering, 22 percent from solutions or sales engineering, 15 percent from data engineering or data science, 10 percent from technical consulting, 8 percent founder or early employee.

One line worth stealing

If your engineers work with AI-generated code, say what proves it correct. Candidates who are good at this role ask that question in the interview, and a description that answers it first signals you have thought about the part that matters.

Ours reads roughly: the code is AI-written and every change passes an eval suite derived from the specification before it reaches the customer's branch, so the engineering judgment goes into the specification and the verification rather than the typing.

Next: how to hire for sourcing and signals, and how to interview for the loop.