Skip to content

About FDEs

What is an FDE (Forward Deployed Engineer)?

An FDE is an engineer embedded at the customer's site, turning AI and software capabilities into systems the customer's business actually uses and that show results.

In one sentence: An FDE (Forward Deployed Engineer) is an engineer embedded at the customer's site, responsible for turning AI and software capability into systems that people actually use in the customer's business and that show measurable results.

What FDEs Do

An FDE's work runs from the customer's business problem to a system that is launched and kept in use. On a typical project, an FDE does the following:

  1. Understand the business: Interview decision-makers, business owners, and IT to clarify the real pain points, data situation, and constraints.
  2. Judge the scenario: Among many possible things to build, pick the one that is worth doing now and can actually be done.
  3. Design the solution: Set the scope and metrics, and state explicitly what is out of scope for the first phase.
  4. Build it: Use models, knowledge bases, tool calls, and system integrations to produce something that works.
  5. Evaluate and verify: Use test sets to prove the system is reliable, and find where it will fail.
  6. Deliver to production: Handle deployment, data compliance, human fallback, monitoring, and rollback, then hand the system over to the customer.
  7. Prove the value: Use the customer's own data to calculate return on investment, and bring problems found on site back to the product team.

How the FDE Role Differs From Other Roles

Role Primarily accountable to Main output Difference from FDE
FDE The customer's business results Systems launched and used on the customer's site —
Algorithm engineer Model performance Models, algorithms FDEs generally do not train models; the focus is making existing models work well
Backend / full-stack engineer Product features Code for a general-purpose product FDEs work on one customer's specific problems
Solutions architect Pre-sales solutions Solution documents, demos FDEs build the solution themselves and own its launch
Implementation consultant Product deployment Configuration and training for a standard product FDEs make judgment calls and do custom development, not just configuration
Product manager Product direction Requirements documents, roadmaps FDEs handle both requirement judgment and engineering implementation

Put simply, the FDE is a role that combines engineering skill, business understanding, and customer communication in one person.

Where the Role Comes From

The term "forward deployed" comes from the U.S. data company Palantir, which sent engineers to customer sites early on to solve problems together with customers. After large models appeared, a large gap opened between model capability and what enterprises could actually use. Many AI companies began creating this role in volume, and demand for FDEs has grown quickly.

Demand and Pay

The following figures come from public reporting. Please check the original sources when citing them:

  • According to recruitment platform data cited by China News Service in a June 2026 report, FDE job postings grew about 42-fold over two years.
  • According to Indeed data cited by China Daily in an August 2026 report, global FDE job postings rose from 643 to 5,330 between April 2025 and April 2026.
  • The same China News Service report listed domestic job postings: ByteDance's Doubao large-model FDE role paid 35,000 to 70,000 RMB per month, with 15 months' salary per year; Ant Digital Technology's FDE role paid 40,000 to 60,000 RMB per month. The report also noted that high pay is concentrated among top talent and does not represent the industry average.
  • China Daily reported that domestic companies are also adopting similar models. For example, Tencent and Changan Automobile formed a joint team of 34 people. The report argued that the domestic FDE talent training system is still close to empty.

Sources: China News Service, 2026-06-05; China Daily, 2026-08-18.

Capabilities an FDE Needs

The FDE Capability Standard published by ChinFDE divides FDE capability into six dimensions:

  1. Business insight and scenario judgment: Distinguish what the customer says they want from what the business actually needs.
  2. Solution design and scope management: Decide what to build first, what not to build, and how to measure success.
  3. Engineering implementation: Assemble models, knowledge, and business rules into a working system.
  4. Evaluation and quality assurance: Prove the system holds up on problems it has never seen.
  5. Launch, delivery, security and compliance: Handle deployment, data compliance, fallback, and rollback.
  6. Communication, collaboration and value proof: Get all parties aligned, and calculate return on investment.

For the full content, see the FDE Capability Standard.

What Is Special About Being an FDE in China

  • Model choice: Most projects use domestic models that can be used in compliance with local rules. You need to understand the characteristics and integration methods of different domestic models.
  • Private (on-prem) deployment: Many enterprises, especially in manufacturing, finance, government, and healthcare, require that data stay inside the intranet. FDEs often have to deliver in environments with no public internet and limited compute.
  • Data compliance: Projects involving personal information must follow personal information protection law. Whether data can leave the country or be processed by third parties must be settled at the solution stage.
  • Wide differences in customers' digital maturity: Some customers have complete systems, while others keep key data in Excel files and paper forms. Judging "what can be done now" matters more than "what is theoretically possible."

Who Suits the FDE Role

  • Backend, full-stack, and algorithm engineers who want to get closer to the business.
  • Technically minded solutions engineers, pre-sales engineers, and implementation consultants.
  • Product managers with a technical background who want to build things themselves.
  • Business professionals with deep knowledge of an industry who are willing to learn AI application development.

What they have in common: they are willing to work with customers, can make judgments with incomplete information, and take responsibility for results.

How to Start Learning

FDE skills are hard to learn by watching videos and answering multiple-choice questions, because the core of the job is making judgments in concrete situations. ChinFDE's approach is to put learners directly into simulated customer projects: interviewing an AI-played customer, writing a solution, building a system and passing hidden tests, and finally producing a launch plan and retrospective. The outputs of the whole process are the basis for assessment.

You can start with the first simulated project.