Ai-Powered-Interview-sysetm

AI-Powered Interview System (NeuroHire)

A Flask + MySQL + Socket.IO web app for conducting live, proctored technical interviews with:

• Scheduling & user management (admin/interviewer vs candidate)
• WebRTC video calling + screen share
• Collaborative code editor (real-time sync)
• Chat (persisted to MySQL)
• Proctoring/anti-cheat via browser-based face/eye detection (face-api.js) + copy/paste blocking
• Resume upload + in-room resume viewer (PDF preview in modal)

---

Table of contents

• Tech stack
• Repository structure
• Interview system features
• Database schema
• Configuration
• Run with Docker (recommended)
• Run without Docker (Windows / XAMPP)
• Usage walkthrough
• Key HTTP routes
• Socket.IO events
• Resume upload & preview details
• Troubleshooting

---

Tech stack

• Backend
  • Python (tested with the repo’s Dockerfile: Python 3.9)
  • Flask
  • Flask-SocketIO + eventlet (realtime)
• Database
  • SQLite (single file, no separate server)
• Frontend
  • Jinja2 templates (server-rendered HTML)
  • Tailwind CSS (CDN)
  • Vanilla JavaScript
  • CodeMirror 5 (editor, CDN)
• Realtime
  • Socket.IO (WebRTC signaling, chat, code sync, join approval, alerts)
• Video
  • WebRTC peer-to-peer with STUN: stun:stun.l.google.com:19302
• Proctoring
  • face-api.js (CDN: @vladmandic/face-api) + local JS logic

---

Repository structure

app.py                  # Flask app + Socket.IO events
config.py               # Env-based configuration
extensions.py           # sqlite helpers + socketio
routes/
  auth.py               # login/logout/setup-admin
  main.py               # dashboard/add-candidate/schedule/interview/resume endpoints
templates/              # Jinja templates (UI)
static/js/              # webrtc/editor/chat/face detection
schema.sql              # Database schema
init_db.py              # Initialize DB from schema.sql
update_db.py            # Apply schema updates (adds missing columns/tables)
update_db_join_status.py# Adds join-status column (older DB support)
uploads/resumes/        # Stored candidate resumes (gitignored recommended)

---

Interview system features

Roles

• Admin / Interviewer
  • Create candidates
  • Schedule interviews (generates a unique meeting link)
  • Join interview room and approve/deny candidate join requests
  • Mark interview as completed
  • View candidate resume inside the interview room
• Candidate
  • Login
  • View scheduled interviews
  • Request to join interview room (waiting-room flow)
  • Participate in interview (video + code + chat)
  • View their own resume inside the interview room

Scheduling & access control

• Session-based auth (Flask sessions)
• Magic meeting link per interview: interviews.meeting_link (UUID)
• Authorization checks: only the scheduled interviewer/candidate can join the room

Live interview room

• Video call
  • Camera + microphone
  • Screen sharing toggle
• Collaborative code editor
  • CodeMirror editor
  • Realtime sync via Socket.IO
  • Last code is persisted in DB (interviews.code_content)
  • Basic anti-cheat: copy/paste disabled in editor
• Chat
  • Realtime messages
  • Persisted in chat_messages
  • Chat history loaded when opening the interview room
• Candidate waiting-room / approval
  • Candidate requests to join
  • Interviewer admits/denies
  • Status persisted in interviews.candidate_join_status
• Resume in interview room
  • Resume uploaded on candidate creation
  • “View Resume” opens a modal in the interview room (PDF preview supported)

Proctoring / cheat detection

• No face detected
• Multiple faces detected
• Looking away (eye landmark heuristic)
• Face too far (bounding-box size heuristic)
• Candidate-side detections trigger cheat-detected → interviewer receives cheat-alert

---

Database schema

The schema is defined in schema.sql. Core tables:

• users
  • role: admin or candidate
  • Candidate profile fields (optional): phone, target_role, experience_level
  • Resume fields (optional): resume_path, resume_original_name
• interviews
  • meeting_link: unique UUID used as the room id
  • code_content: last saved editor content
  • candidate_join_status: pending/requested/approved/rejected
• chat_messages
  • Linked to an interview via interview_id

---

Configuration

Environment variables are loaded via python-dotenv (.env) and read in config.py.

Required/commonly used

• SECRET_KEY: Flask session secret

Optional

• DATABASE_PATH: Full path to SQLite database file. If not set, uses instance/interview_system.db in the project directory.
• INSTANCE_PATH: Directory for the DB file (used if DATABASE_PATH is not set).

Upload limit

• MAX_CONTENT_LENGTH: max request size in bytes (default: 10MB)

---

Run with Docker (recommended)

1) Start containers

docker compose up --build

2) Open the app

• App: http://127.0.0.1:5000

The app uses SQLite by default; the database file is created on first run (python init_db.py or when the app starts with an existing schema).

---

Run locally

1) Create & activate a virtualenv (optional but recommended)

python -m venv .venv
.\.venv\Scripts\activate

2) Install dependencies

pip install -r requirements.txt

3) Configure .env (optional)

Set SECRET_KEY. Optionally set DATABASE_PATH to a custom SQLite file path (default: instance/interview_system.db).

4) Initialize the DB

python init_db.py

If you already created the DB earlier and need new columns/tables:

python update_db.py
python update_db_join_status.py

5) Run the server

python app.py

The server is configured to bind to 0.0.0.0:5000 (LAN/Docker friendly).

---

Run without Docker (Windows)

Prerequisites

• Python 3.9+ installed and available in PATH (python --version)
• Pip available (pip --version)
• No separate database server needed (SQLite is file-based)

1) Create & activate a virtual environment

cd c:\xampp1\htdocs\Ai-Powered-Interview-sysetm
python -m venv .venv
.\.venv\Scripts\Activate.ps1

2) Install dependencies

pip install -r requirements.txt

3) Configure .env (optional)

Create/update .env in the project root. Example:

SECRET_KEY=dev
# Optional: custom SQLite path (default: instance/interview_system.db)
# DATABASE_PATH=./data/interview_system.db
MAX_CONTENT_LENGTH=10485760

4) Initialize the database schema

python init_db.py

If you already had a DB and need to add the newer columns/tables:

python update_db.py
python update_db_join_status.py

5) Run the server

python app.py

Open:

• http://127.0.0.1:5000

Note: the server binds to 0.0.0.0, so you can also access it from another device on your LAN using your PC’s IP (e.g. http://192.168.x.x:5000).

---

Usage walkthrough

1) Create the first admin

• Visit: /setup-admin
• Create an admin account

2) Login

• Visit: /login

3) Add a candidate (with resume)

• Visit: /add-candidate
• Upload a PDF/DOC/DOCX resume (optional)
• Submit the form

Resumes are stored on disk under uploads/resumes/ and the path is stored in MySQL.

4) Schedule an interview

• Visit: /schedule-interview
• Pick the candidate and set date/time
• A meeting_link UUID is generated and stored in interviews.meeting_link

5) Join the interview room

• From dashboard, click Join
• URL format:
  • /interview/<meeting_link>

6) Candidate join approval (waiting room)

• Candidate clicks “Ask to Join”
• Interviewer receives a popup to Admit or Deny

7) View resume in the interview room

• Click View Resume
  • PDF: previews in a modal iframe
  • DOC/DOCX: offers an “Open resume” link (browser download/open behavior)

---

Key HTTP routes

Auth

• GET/POST /login
• GET/POST /setup-admin
• GET /logout

App

• GET /dashboard
• GET/POST /add-candidate
• GET/POST /schedule-interview
• GET /interview/<meeting_link>
• POST /interview/<meeting_link>/complete
• GET /interview/<meeting_link>/resume (authorized users only)

---

Socket.IO events

Room membership / join approval

• Client → Server
  • join { room, username }
  • leave { room, username }
  • request-join { room, username } (candidate)
  • approve-join { room } (interviewer)
  • reject-join { room } (interviewer)
• Server → Client
  • joined
  • user-joined, user-left
  • join-request
  • join-approved
  • join-rejected
  • candidate-approved

WebRTC signaling

• offer, answer, ice-candidate

Collaborative editor

• code-change { room, code } (also persists to interviews.code_content)
• code-update { code }

Chat

• chat-message { room, message, username, timestamp } (also persists to chat_messages)

Proctoring alerts

• cheat-detected { room, type, message, timestamp }
• cheat-alert { type, message, timestamp } (server broadcasts to interviewer)

---

Resume upload & preview details

• Upload accepted types: PDF, DOC, DOCX
• Max size: 10MB
• Stored on disk: uploads/resumes/<uuid>_<originalname>
• DB fields (in users):
  • resume_path
  • resume_original_name

Resume serving endpoint is scoped to the interview:

• GET /interview/<meeting_link>/resume
  • Only the interview’s interviewer/candidate may access it

---

Troubleshooting

Database issues

• Run python init_db.py to create the SQLite database and tables.
• If the DB file is in a read-only or missing directory, set DATABASE_PATH in .env to a writable path.

Resume button not visible in interview room

• The candidate must have a stored resume (users.resume_path not NULL)
• Re-upload the resume via Add Candidate if needed

PDF preview is blank

• Check the resume endpoint opens directly in browser:
  • /interview/<meeting_link>/resume
• Some browsers/extensions block iframes; try “Open resume” link.

WebRTC video not connecting

• Both peers must allow camera/mic permissions
• If behind strict NAT/firewall, you may need a TURN server (STUN-only is not always enough)

---

Security notes (recommended improvements)

• Store uploads outside the repo and add uploads/ to .gitignore.
• Consider virus scanning for uploaded files (production).
• Add stricter MIME validation (not only file extension).
• Add CSRF protection for forms (Flask-WTF).
• Use HTTPS in production (required for many WebRTC features).

---

License

Add a license if you plan to publish/distribute this project.