FDE

FDE stands for Forward Deployed Engineer: the engineer who goes inside a customer organization and makes their own company's product actually run there. You work for the product company, and you write production code on the customer's ground. Palantir created the role in 2011. Today it is hired mostly by companies running enterprise AI transformations.

What the work is

Five things run in order, and then run again, inside a single engagement. In OpenAI's words for the role, you "own discovery, technical scoping, system design, build, and production launch."

1. You define the problem. You sit with the customer's engineering team and their operators to work out which workflows to automate, how far to go, and what data exists. You do not receive a requirements document. You see the ground first, then decide what is needed.
2. You build a prototype and integrate it. You write the code that connects your product to the customer's data, infrastructure, and permission model. Frontend or backend, whatever it takes. How much you code varies by company, and it moves through the phases of a deployment.
3. You take it to production and keep it stable. You investigate failures that only happen in the customer's environment, find the cause, ship the fix, and monitor it. Palantir describes this as coordinating between the product team, the deployment team, and the customer while protecting reliability.
4. You talk to both sides. You negotiate scope and priority between the customer's stakeholders and your own product, research, and sales teams.
5. You bring the field back home. Patterns that repeat across customers become tooling, playbooks, and product features. Palantir has said that some of its most valuable product additions started in the field.

How it differs from staff augmentation

"Embedded with the customer" sounds like contract staffing. It is not. Who employs you, who directs you, what you own, and how much time you spend on site are all different.

• Employment and direction are different. A staff-augmentation contractor is placed with the client and directed by the client. An FDE is a full-time employee of the software company. Direction stays with your employer, even while you are on the customer's site.
• The job is different. A contractor executes requirements that were written before they arrived. An FDE starts by defining the problem alongside the customer and designing how the product gets adopted.
• It is not full-time on site. Palantir engineers are understood to spend about 25% of their time at customer sites. OpenAI's Korea posting states roughly 50% travel.

How it differs from nearby roles

• Solutions Engineer: different point in the cycle. A Solutions Engineer builds the demo and the proof of concept before the contract, then moves to the next customer. An FDE arrives after the contract and stays until the product runs in the customer's environment.
• Sales Engineer: different function. A Sales Engineer sits in the sales organization and is measured on closing. An FDE is an engineering role. When Bloomberry analyzed 1,000 FDE postings, not one carried a revenue quota.
• Backend Engineer: opposite direction. A Backend Engineer builds one feature that many customers use. An FDE builds many features for one customer. Both write code.

What companies look for

Bloomberry's analysis found that the most common prior role in FDE postings was backend or full-stack software engineer at 45%, followed by solutions engineers and architects, then data engineers. The wording varies by company, but the requirements converge. Lay the public postings from global and Korean AI companies on top of each other and three things are always there.

• Production-level engineering. You can write and review frontend and backend code in Python, TypeScript, or similar. This is a customer-facing role, but it is not a low-code one.
• Experience working directly with customers. Scoping and prioritizing while the requirements are still vague. Consulting, solutions engineering, or backend work taken straight from customer requests all count.
• Experience putting an LLM into a real system. Not training models. Building or deploying model-based systems, and understanding how model behavior shows up in the product experience.

FAQ

Is FDE a good career move? Demand is rising. Having taken a product all the way to working inside a customer's environment is rare anywhere. But the same title can mean very different work. At some companies it sits closer to presales or customer support than to engineering. Check whether you would be writing production code, and which organization the role reports into, before you decide.

Are companies hiring FDEs in Korea? Yes. When a global AI company enters Korea or APAC, FDE is one of the first roles they open, often before the sales organization. Someone has to build the product into local customer environments first. Korean AI companies and large IT services firms have started building teams under the same name.

What does an FDE earn? There are published surveys abroad. In Korea the role is new enough that public salary data is limited. The spread between Korea-based roles at global AI companies and roles at domestic companies is wide enough that a single range would not mean much.