What is a forward deployed engineer? A plain definition
A forward deployed engineer, often shortened to FDE, is an engineer who works inside the customer’s company. They use the customer’s real data and the customer’s real processes, and they build the solution there. A product company ships the same software to many customers from a distance. A forward deployed engineer builds one thing for one customer, on site or in close contact with the team.
Where the term comes from
Palantir popularised the title. It sent engineers to its customers to make its software work on their data. The idea has since moved to AI projects, where the hard part is rarely the model. The hard part is the messy reality around it: documents in five places, a process that lives in someone’s head, exceptions nobody wrote down.
What a forward deployed engineer does
- Sits with the people who do the work and watches how it is done today.
- Reads the real documents, mails and data, not a cleaned-up sample.
- Builds a thin but working version and puts it in production early.
- Collects feedback from the users and changes the system until it does the job.
- Hands over something the team can keep using, with a named contact for what comes next.
How it differs from nearby roles
| Role | Works on | Main output |
|---|---|---|
| Consultant | The strategy and the organisation | Analysis, recommendations, a roadmap |
| Solutions engineer | Integrating a vendor’s product | A demo or an integration of that product |
| Software engineer at a product company | One product for many customers | A product feature |
| Forward deployed engineer | One customer’s data and processes | A working system for that customer |
When a company needs one
Three situations come up most often. The process is manual and nobody has had the time to document it. The data sits in several tools and no off-the-shelf product reads all of it. A first pilot was built and stalled before anyone used it. Add a short deadline and no in-house machine learning team, and the role fits well.
It does not fit everything. If a standard tool already covers the need, such as accounting or a CRM, buy the tool.
What good looks like
A good forward deployed engagement shows something running early, so that real users can react to it instead of a slide deck. People validate the output before anything leaves the company. A person on the customer side owns the result.
This is how we work at Saména, the agency that publishes this blog. After a first 30-minute call, a first agent runs two weeks later, and the team validates what it prepares before anything is sent. The same loop applies to machine learning and data projects, not only to agents.
Questions
What does a forward deployed engineer do?
They work inside the customer's company, read its real documents and data, build a working version of a solution, put it in front of the people who will use it, and adjust it on their feedback until it does the job.
Is a forward deployed engineer the same as a consultant?
No. A consultant's main output is analysis and recommendations. A forward deployed engineer's main output is a working system running on the customer's own data.
Does a company need a machine learning team to use one?
No. The engineer brings the technical skills. The company needs a person who knows the process and can decide what is good enough to go live.
How long does a forward deployed project take to show a result?
It depends on the project. At Saména, a first agent runs two weeks after the first call, and it is then adjusted on the client's feedback.