EvalcueAI

EvalcueAI is a full-stack platform for software engineering interview practice and structured technical hiring. Candidates can rehearse conversational, coding, and system-design interviews with adaptive AI follow-ups. Hiring teams can create role-specific assessments, including multi-file debugging assignments, invite candidates, collect technical evidence, and review scorecards while keeping employment decisions human-controlled.

Product surfaces

EvalcueAI is deployed as three intentionally separated product surfaces from the same codebase. VITE_APP_SURFACE and deployment guards prevent Practice and Hiring from becoming path-only variants of one application.

EvalcueAI Landing

• Product overview and public documentation
• Entry points into the separate Practice and Hiring products
• SEO/indexable marketing and resource pages

Production host: evalcueai.com

Canonical public route in the shared client: /

EvalcueAI Practice

• Role- and job-description-specific interview plans
• Conversational interviews with hands-free voice support
• Coding / online-assessment rounds with sandboxed code execution (Node.js 22, Python 3.12, Java 21, C++20)
• Live system-design discussions with an Excalidraw architecture canvas
• Adaptive questioning based on answer evidence, competency coverage, and resume context
• Resume upload, review, JD matching, JD-tailored one-page resume generation, saved interview experiences, progress tracking, and reminders
• Post-interview feedback and improvement suggestions

Production host: practice.evalcueai.com

Canonical public route in the shared client: /practice

EvalcueAI Hire

• Organization-owned technical assessments and candidate pipelines
• Manual or AI-generated round definitions
• Conversational, coding, live system-design, and multi-file debugging assessment modes
• Debugging assignments support either source-code fixes evaluated by recruiter-authored hidden tests with candidate-safe result names, or file-scoped findings-only diagnosis
• Invite-only candidate links, invitation lifecycle tracking, and local answer recovery
• Weighted competency scorecards, AI evaluation, human overrides, and calibration views
• Optional consented integrity signals such as fullscreen/focus/clipboard/connectivity events and on-device face-presence checks
• Fairness report: scoring and selection rates, impact ratios and four-fifths flags by sex, race/ethnicity and intersectional group, built from voluntary post-submission self-identification (groups under 5 suppressed; an input to, not a substitute for, an independent bias audit)
• OIDC SSO support for eligible hiring organizations

Production host: hiring.evalcueai.com

Canonical public route in the shared client: /hire

Architecture

React 19 / Vite / MUI
        |
        v
Express 5 API
  |       |        |          |
MongoDB  Redis    AI APIs    External services
         BullMQ   OpenAI /   PayU, Brevo,
                  Gemini     Cloudinary, code runner

• Client: React, React Router, Material UI, Axios, Excalidraw, Monaco
• API: Express, Zod, Mongoose, JWT-based auth, organization authorization
• Data: MongoDB; production transaction support is required
• Async work: Redis + BullMQ for question preparation, bulk feedback, and candidate evaluation
• AI: OpenAI and/or Gemini, with optional Tavily grounding
• Code execution: a self-hosted runner service (runner/: Node.js HTTP API + nsjail sandbox) for single-file runs and multi-file debugging projects
• Operations: Prometheus metrics, optional Grafana OTLP push, Sentry, structured logs

Repository layout

client/                  React/Vite SPA
  src/components/        shared UI, interview workspaces, SEO helpers
  src/pages/             public, practice, hiring, and admin pages
  e2e/                   Playwright browser journeys
server/                  Express API
  src/controllers/       request/domain controllers
  src/models/            Mongoose models
  src/queues/            BullMQ queues and workers
  src/routes/            API routing and authorization composition
  src/services/          business logic and integrations
  src/test/              unit and API journey coverage
observability/           dashboards, PromQL, and alert guidance

Getting started

Prerequisites

For local development: Node.js 22 (see .nvmrc; nvm use), MongoDB, and credentials for whichever integrations you want to exercise. AI features require OpenAI or Gemini. Resume storage requires Cloudinary.

Production additionally requires Redis, HTTPS, Brevo transactional email, CAPTCHA, PayU (including Zion for automatic renewal), MongoDB transactions, and any dependencies for enabled STT/code-execution features.

Install

cd server && npm install
cd ../client && npm install

Configure the server

The checked-in example contains development-safe defaults rather than production settings:

cp server/.env.example server/.env

At minimum, set a real JWT_SECRET and a reachable MONGO_URI. Configure OPENAI_API_KEY or GEMINI_API_KEY to use AI features. PRACTICE_CLIENT_ORIGIN and HIRING_CLIENT_ORIGIN are the canonical product browser origins; CLIENT_ORIGIN remains the local/legacy fallback.

For production set NODE_ENV=production and configure, at minimum:

This is the beginning of the README. Read the rest at the original. Read the full README ↗