Skip to main content

Glossary

Terms you'll hear in week one, grouped by area. One line each.

People and roles​

  • Candidate (B2C) — a job seeker; takes AI interviews, optimizes their resume, builds a Profile Card.
  • Employer / Recruiter (B2B) — a company hiring through the platform; creates job postings, runs interviews, reviews transcripts, summaries, and graded candidates.
  • Self-serve user — an individual using the platform on their own, primarily for practice interviews not tied to an employer posting.
  • Professional interviewer — a vetted human interviewer who conducts paid sessions, sets availability, and requires admin approval.
  • Admin — platform administrator; manages users, sessions, analytics, content, and platform configuration.

Interviews and sessions​

  • AI-bot interview — the flagship voice interview: a spoken session with the AI interviewer, with real-time transcription, evaluation, and feedback.
  • AI mock interview — the text-based interview format: typed Q&A with AI evaluation and scoring.
  • Practice vs. lifelike interview — practice sessions use a lightweight local interviewer face; lifelike (paid) sessions use the live photo-real avatar.
  • Blended interview — a technical interview that interleaves theoretical questions with short practical coding exercises in one session, with the code editor appearing only on coding questions.
  • Coding interview / code evaluation — interviews where the candidate writes code in an in-browser editor; code answers are scored alongside spoken answers.
  • Session room — the in-call interview UI where a live session actually runs (a dedicated app embedded by the main web app).
  • Interview room — the page in the main web app that hosts a live session (AI interview room or human meeting room).
  • Heartbeat / zombie session cleanup — liveness signal from an active session; sessions abandoned mid-interview are detected and credited back rather than silently consumed.

Hiring flow​

  • Job posting — the employer-side object a candidate interviews against; carries the role context and session configuration such as the recording policy.
  • JD funnel — the recruiter flow for creating a job posting from a job description.
  • Recording policy (OFF / OPTIONAL / MANDATORY) — per-posting setting governing session recording: not recorded; candidate-controlled with a consent notice; or automatic with a one-time acknowledgment. Self-serve practice defaults to OPTIONAL.
  • Profile Card — a shareable, credibility-graded summary of a candidate, with scores computed from completed interviews; generated on demand and exportable.
  • Feedback report — the post-interview output: per-answer evaluation, ratings, written feedback, transcript, and any recording; exportable as PDF.
  • ATS score — the AI resume optimizer's score of how well a resume parses and matches a role (resume-to-job match analysis).

Plans and billing​

  • Plan — a subscription tier (free or paid, for individuals or companies), purchased through Stripe-powered checkout.
  • Entitlement — what a plan allows: which features a user can access and how much usage (e.g. AI interviews, AI feedback) they can consume.
  • Usage / subscription metrics — the running count of consumed entitlements a user can see against their plan limits.
  • Credit-back — returning a consumed interview entitlement when a session fails through no fault of the user (e.g. avatar failure, abandoned session).

Real-time platform​

  • LiveKit — the open-source WebRTC stack the platform uses for real-time audio/video.
  • SFU (Selective Forwarding Unit) — the media server model LiveKit uses: each participant sends one stream up, and the server forwards streams to the others.
  • Room / access token — a short-lived credential minted by the backend that authorizes a specific user into a specific session room.
  • Egress — LiveKit's server-side recording pipeline: the session is recorded on the server and the file is uploaded to cloud storage.
  • STT (speech-to-text) — real-time transcription of the candidate's spoken answers.
  • TTS (text-to-speech) — neural voice synthesis that speaks the AI interviewer's questions aloud.
  • Avatar — the photo-real animated interviewer face, lip-synced to the TTS audio.
  • Orb fallback — the graceful-degradation visual shown when the avatar can't render on the candidate's device; the interview continues with audio and subtitles.

Ways of working​

  • Epic / story — an epic is a larger scoped outcome; stories are its smallest shippable units. Stories deploy to dev incrementally; production is promoted once per epic.
  • TI-XXX — a Jira ticket key; every task has one, and commit messages start with it so work links back to the ticket.
  • Feature flag — a configuration switch that lets new behavior ship disabled and be turned on (or rolled back) without a redeploy.
  • Entitlement check — the single shared mechanism for asking "can this user do this on their plan?" — never hardcoded per-plan logic.