>_Interview Orchestrator
AI-assisted interview orchestration

Run coding interviews with sandboxed sessions, traceable AI collaboration, and calibrated judges.

Provision a sandbox, send a candidate link, and get back a scored report backed by hidden tests, commit evidence, and Phoenix-traced agent calls. No spreadsheets, no guessing — just the signal.

interviewer access only

The session lifecycle

Click through each stage to see what happens behind the scenes.

step 1 / 6create

Create session

Interviewer configures a task, selects judges, and sets difficulty.

The Orchestrator issues a session_id and a candidate launch link.

What you'll see

Three surfaces — interviewer dashboard, candidate IDE, and evaluation report.

interview-orchestrator — preview
Active
1
Provisioning
1
Evaluating
1
Completed
1
CandidateTaskStatus
Candidate ATodo REST APIcompleted
Candidate BReact Component Challengeactive
Candidate CTodo REST APIevaluating
Candidate DTodo REST APIprovisioning

Built for engineering teams

Sandboxed IDE

Each candidate gets an isolated E2B sandbox with code-server, tests, and a scoped gateway key.

Frozen commits

Submission freezes a commit SHA — the exact code bundle that judges evaluate, no ambiguity.

Independent judges

AI Collaboration, Code Quality, and Test Coverage score on a calibrated 0–3 scale.

Traceable evidence

Every report cites Phoenix trace spans and commit SHAs — claims link back to real signal.

Ready to run an interview?

Log in to create a session and send a candidate link.