Forward deployed engineer interviews that test the code and the customer

A forward deployed engineer, or FDE, writes code and scopes problems with the customer in the same week. Codility tests both halves in three real-world tasks: scope the customer’s request, build with Claude Code or Codex, then write the reply to the customer.

AI tools included, no API keys needed Validated assessments since 2009 SOC 2 Type II ISO 27001

Trusted by teams at

Tesla Accenture Google Deloitte BMW Group SpaceX Deutsche Bank GitHub Bain & Company American Express McLaren Racing LSEG Cockroach Labs Microsoft KPMG Halliburton EY Barclays Skyscanner ABB Vinted HSBC PwC Palo Alto Networks TD Broad Institute Publicis Groupe Zalando ING NIQ Nissan Korn Ferry

Most FDE interview processes test only half of the job

01

A coding test shows how someone writes code. A case study shows how they structure a business problem. Neither shows whether the same person can do both with a real customer in the same week.

02

Every company builds its own interview process from take-homes, customer role-plays, case studies and interviews where the candidate breaks a vague problem into a plan. Few are scored the same way from one interviewer to the next.

03

FDEs ship with coding agents, so a take-home without AI tests a job that no longer exists. With AI allowed, the finished code shows little about how the candidate directed the agent.

An FDE’s job runs before, during and after the code

In open FDE job postings, 92% describe building and shipping production software. 78% describe scoping problems with the customer, and 66% make customer-facing experience a hard requirement. The usual tests each see only part of that.

One assessment that follows the job: plan, build, communicate

Plan: scope the customer’s request

The candidate starts from an incomplete brief and questions simulated stakeholders, such as a CTO, a business owner and procurement, to find out what the customer needs. Each stakeholder holds facts that are missing from the brief, and the score reflects how much of that information the candidate draws out.

Codility task workspace with the task brief and an open source file on the left and the AI Assistant panel on the right, sample data.
Codility engineering assessment task in a VS Code style workspace with the code editor, a failing test in the terminal and the AI Assistant panel, sample data.

Build: write the code with Claude Code or Codex

The candidate builds a system in Python that sorts incoming customer emails and routes each one to the right workflow. They use the real Claude Code or Codex when you allow it, with AI usage included and no API keys to set up. Hidden tests score the result.

Communicate: write the reply to the customer

The candidate reads a new request from the customer and decides whether it refines the agreed scope or expands it. Then they write the reply: what changes, what it costs in scope, speed and quality, and what the customer has to decide.

A candidate's submitted written response beside the saved AI Assistant conversation, sample data.
Assessment report showing the candidate's submitted code on the left and the saved AI Assistant and Claude Code conversation on the right, sample data.

See both halves of the job in one report

Each task gets its own score, so a strong coder who skipped the scoping stands out. A full assessment takes 60 to 90 minutes for one role, and every candidate gets the same tasks and the same scoring rules. Codility gives your hiring team the evidence, and your team makes the final decision.

  • A separate score for Plan, Build and Communicate
  • Every prompt to Claude Code or Codex, saved to the report
  • Who directs the agent and who accepts its first answer

From the first screen to the engineers you already have

Screen runs the three tasks at volume before the interview, and every AI interaction is saved as reviewable AI activity.

Interview takes the final round live in a shared VS Code environment with Claude Code and Codex, a whiteboard and a full transcript. When the role needs it, add a business task to see how the candidate works with AI outside the code. Your interviewer runs the customer conversation and judges it.

Skills Intelligence runs the same assessment across your current engineers, so you can see who is ready to move into an FDE role and what each one still needs to learn.

See agentic coding interviews

Built on almost two decades of assessment science

1,300+validated tasks in 80+ languages and frameworks
87%of candidates rate their Codility experience good or excellent
91%of candidates rate the assessment content fair

Every FDE you hire sits in front of a customer.
Make decisions you can explain and defend.

Candidate figures: Codility candidate feedback survey, October 2025 to April 2026, across Codility assessments. 46,863 responses for the 87% figure and 45,925 for the 91% figure.

Bring your FDE job description

The assessment is ready-made, then tuned to your role. Companies define the FDE role differently, so the assessment science team sets it up with you in four steps.

Share the role

Send the job description, and tell us what separates a strong FDE at your company from an average one.

See what each task measures

The assessment science team maps your role to Codility’s skills and shows you how each task is scored.

Choose Screen, Interview or both

Screen runs the assessment at volume before the interview. Interview takes the final round live.

Calibrate with your own engineers

Your hiring managers and best FDEs take the tasks first, so the passing score reflects people who succeed in the role.

What happens next

  1. Tell us about the role in the form.
  2. A specialist shows you the three tasks, how each one is scored and how candidates work with Claude Code and Codex.
  3. The assessment science team tunes the tasks to your job description, ready in as little as a week.

Talk to us

Before you interview forward deployed engineers