Skip to content

Practice at a simulated customer site,
and become a Forward Deployed Engineer

ChinFDE is a learning and certification platform for Forward Deployed Engineers (FDE), focused on applied AI delivery. Build your foundation with the structured course, with an AI tutor in every lesson. Then take on simulated customer projects: interview an AI-played customer, define the scope, build the system and pass the tests. Finish the projects and your competency certification is complete.

Learning and certification at L1 Foundation are currently free.

Requirements interview · Example

A logistics company · Customer service manager

AI customer
The boss wants an AI that can answer any question. From your side, what takes up the most time in customer service right now?
I haven't thought about “any question.” Honestly, about 80% of calls are asking the same thing: where is my shipment?
How is that handled today? How long does one case take?
Agents look it up by hand in the waybill system, two or three minutes per case. That system has no external API.
The customer says “answer anything,” but only one thing really eats up staff time. Whether you find it depends on how you ask.

What is an FDE (Forward Deployed Engineer)?

FDE stands for Forward Deployed Engineer. FDEs are embedded at the customer's site and turn AI and software capabilities into systems the customer's business actually uses. The work runs from requirements interviews and solution design through hands-on building, launch and delivery, to proving the results.

FDE work and competency requirements →

What is ChinFDE?

ChinFDE is a learning and certification platform for FDEs. We publish the open FDE Competency Standard, provide simulated delivery projects in which AI plays the customer, and issue publicly verifiable certificates based on what learners actually produce in those projects.

Read the FDE Competency Standard →

For engineers with 1–3 years of experience: 13 chapters, 9 labs, and an assistant on call

Designed for engineers who have worked in backend, implementation or pre-sales: customer data integration, permissions and network troubleshooting, on-premises and Xinchuang (domestic IT) deployment, then RAG, agents and evaluation-driven delivery. Every chapter shows how the material is used at a customer site. Complete beginners can start with the preparatory track.

  • Stuck? Ask. The tutor knows which lesson you're on. Select any text to ask about it.
  • Self-check questions have no options. Write your understanding and the tutor tells you what you missed.
  • Labs run on your own computer. Paste your records and check them against the acceptance criteria line by line.
View the course outline
  1. 6Field delivery and business judgmentWeeks 17–20

    Requirement breakdown · Stakeholders · Change requests · Ontology modeling · Thin slices · Runbooks

  2. 5Applied AI engineeringWeeks 13–16

    RAG · Agents · Evaluation loops · On-premises and Xinchuang deployment · Chinese models

  3. 4Automation and observabilityWeeks 11–12

    Docker · OpenTofu · CI/CD · Logs, metrics and traces · Cost

  4. 3Cross-cloud infrastructure and securityWeeks 7–10

    Primitives across four clouds · IAM · VPC · Storage choices · Secrets

  5. 2Programming and data integrationWeeks 3–6

    Python · SQL · Integration engineering · Dirty data cleaning · Reconciliation

  6. 1Computer and network fundamentalsWeeks 1–2

    Linux · File permissions · SSH · IP/DNS/HTTP · Git · Environment variables

Learn from the bottom up. Every layer leaves a verifiable output.

Learning is the exam: one project, four stages, evidence at every step

FDE skill shows up as judgment in specific situations, so we don't write test questions. We give you one customer. Every output you leave in the project is the basis for scoring.

  1. 01

    Requirements interview

    Talk with the AI-played boss, business lead and IT. Each has a different stake, and none of them volunteers the full picture.

    Output: interview notes

  2. 02

    Proposal and scope

    Decide which use case to pursue. Write down what you will do, what you won't do, and which metric proves it works.

    Output: proposal document

  3. 03

    Build and evaluate

    Write the system prompt, pick knowledge from a mix of old and new material, then pass a set of tests you cannot see.

    Output: system configuration and test results

  4. 04

    Launch and retrospective

    The boss changes their mind at the last minute and you must respond. Write the rollout, fallback, compliance and ROI plans.

    Output: launch and retrospective documents

Three AIs, each with its own job, none sharing information

The assessment allows AI tools by default, because that is how real FDEs work. So it tests judgment: which questions to ask, which use case to pick, which requirements to cut, and how to prove the solution works.

AI customer
Plays every stakeholder in the project. It answers as specifically as you ask. Information you don't ask for stays out of reach, just as on a real site.
AI tutor
Thinks alongside you throughout. It asks follow-up questions, suggests angles and teaches methods, but does not answer for you, and it does not know what the customer is hiding.
AI examiner
After you submit, scores each of the six dimensions against the public standard, with reasons and suggested next steps.

Three industries, three typical AI adoption challenges

View all projects →
Project 1Healthcare Services · Multi-Location ClinicsBeginner

Yuechi Dental: A Consultation Assistant for a Dental Clinic Chain

The boss wants a “fully automatic AI customer service.” Your job is to find what really needs to be built.

A dental clinic chain with 12 locations wants an AI customer service system. You need to interview three stakeholders with different positions, decide what to build and what not to build, configure a consultation assistant that passes the hidden test set, and deliver a go-live proposal.

~90 minFocus: Problem framing, Design and scope, Evals and quality
Project 2Manufacturing · Injection MoldingIntermediate

Hengrui Precision: An Equipment Maintenance Assistant for the Shop Floor

The client wants “AI predictive maintenance,” but the data cannot support that goal

An injection molding plant wants to use AI for predictive maintenance. In your interviews, you need to uncover the gap between the current state of the data and the goal, narrow the project to a scenario that can be delivered now, and configure a maintenance assistant that never compromises on safety.

~100 minFocus: Problem framing, Engineering, Delivery and compliance
Project 3Manufacturing Export · B2B TradeChallenge

Haituo Pump Industry: An Export Inquiry Triage and Reply Assistant

The boss wants “AI to reply and quote automatically.” Your job is to protect the human review step.

A pump exporter is responding to inquiries too slowly, and the boss wants AI to reply and quote fully automatically. You need to design a human-in-the-loop solution and configure an assistant that triages inquiries, drafts English replies, and never discloses floor prices or overstates certifications.

~110 minFocus: Design and scope, Engineering, Communication and value

Six dimensions that define a qualified FDE

The FDE Competency Standard v0.1 is published under CC BY 4.0. Every score on the platform maps to a dimension in the standard.

Read the full standard →
D1Business insight and problem framing
Is what the customer asks for the same as what the business actually needs?
D2Solution design and scope management
What goes first, what is explicitly out of scope, and which metric proves it works?
D3Engineering implementation
Can they assemble models, knowledge and business rules into a system that works?
D4Evaluation and quality assurance
How do you prove it holds up on questions it has never seen?
D5Delivery, security and compliance
Can the data leave the network, who catches errors, how do you roll out gradually and roll back?
D6Communication and proving value
Can they get the boss, the business and IT all willing to move forward, and make the numbers add up?

Every certificate can be publicly verified

L1 Foundation

Open · Free

Pass all simulated projects. The AI examiner scores your work, and the hidden test set gives an objective result. Proves foundational competence and learning achievement.

Each certificate has a unique number. Anyone can check it on thecertificate verificationpage.

Start learning free
L2 ProfessionalNot open yet
A timed, comprehensive project. Submit a working system and defend it under real-name verification. AI scoring plus human review. Paid once open.
L3 SeniorNot open yet
Submit a real project case for review by a panel. Designed for senior FDEs who lead complex projects.

To learn what a certificate does and does not mean, see thecertification overview.

Hiring FDEs? Replace the written test with a simulated delivery

Candidates and an AI-played customer go through interview, proposal, build and testing. You get a six-dimension scoring report, the full interview record and the hidden test results. Learning is free for individuals. Enterprise assessments are billed per assessment or annually, and the first seed companies get a free pilot.

Before you start: common questions

All questions →
What is an FDE?

FDE stands for Forward Deployed Engineer. FDEs are embedded at the customer's site and turn AI and software capabilities into systems the customer's business actually uses. The work runs from requirements interviews and solution design through hands-on building, launch and delivery, to proving the results.

What is ChinFDE?

ChinFDE is a learning and certification platform for FDEs. We publish the open FDE Competency Standard, provide simulated delivery projects in which AI plays the customer, and issue publicly verifiable certificates based on what learners actually produce in those projects.

What does "learning is the exam" mean?

There are no video lectures and no multiple-choice questions. Each learning unit is a simulated customer project: you interview the AI-played customer, write a proposal, configure the system and face hidden tests, then write a launch plan and a retrospective. The interview notes, documents and test results you produce along the way are the basis for scoring.

Can I use AI tools while learning?

Yes. Real FDEs use AI tools every day, so the assessment does not test memorization or hand-written code. It tests judgment: which questions to ask, which use case to pick, which requirements to cut, and how to prove a solution works. You have to make those judgments yourself.

What background do I need?

The L1 Foundation level does not require coding, but you should understand the basics of LLM applications, such as what prompts and knowledge bases are. Experience in software development, product, implementation or pre-sales makes it easier to get started.

Your first client is already waiting

Mr. Wu, CEO of the simulated client Yuechi Dental, wants a “fully automated AI customer service.” Go talk to him and find out what you should really build.

Start learning free