Skip to content
BranchesTags

Repository files navigation

ELEVATR — AI Hiring Platform

Unified AI-powered platform for resume analysis, resume building, JD-driven pipeline creation, secure video interviews, and generated HR/candidate reports.

Team: Ayush Mayekar · Aaryan Gole · Dnyanesh Panchal · Vaibhav Narute · Tejashree Karekar · Charmi Jani — VCET
Core stack: Next.js · Vite/React · FastAPI · Flask · TypeScript · Python · Supabase · Google Gemini

What ELEVATR Does

ELEVATR connects candidate tools, HR workflows, interview automation, and PDF reporting into one system.

Module Purpose Main Stack
Candidate portal View opportunities, apply to roles, track status, and open interview links Next.js App Router, React, Supabase, TypeScript
HR dashboard Create and manage job pipelines, review candidates, send interview invitations, and view analytics Next.js App Router, Supabase, TypeScript, server API routes
Resume analysis Parse resumes, compare against JDs, score matches, surface skill gaps, and generate feedback Vite/React, FastAPI, Gemini, PyMuPDF, PyPDF2, spaCy, scikit-learn
Resume builder Build/export CVs from templates and candidate data Vite/React, Python backend, Jinja2, reportlab, fpdf2, LaTeX templates
Job pipeline creation Create JD-based pipelines, shortlist candidates, and feed interview scheduling Next.js API routes, Supabase tables, Zod validation
Video interview Run secure invite-based AI interviews with speech input/output Flask, Gradio client, gTTS, speech_recognition, FFmpeg, Gemini
Reports Generate HR and candidate-safe PDFs for resume analysis, interview results, and summaries pdf-lib, reportlab, fpdf2, Jinja2

Tech Stack By Module

1. Candidate Portal

  • Built with Next.js 14, React 18, and TypeScript.
  • Uses Supabase auth/session handling for login, signup, and protected routes.
  • Candidate dashboard reads opportunities and application status from API routes.

2. HR Dashboard

  • Built in Next.js with server-side API routes under app/api/*.
  • Uses Supabase for pipelines, applications, interviews, and analytics.
  • Handles batch screening, invite generation, reporting links, and status updates.

3. Resume Analysis

  • Frontend: Vite + React application in Client Side/ai-skill-analyzer-main.
  • Backend: FastAPI service in Client Side/backend.
  • AI: Google Gemini, with supporting NLP/document tooling for scoring and extraction.
  • Libraries: PyMuPDF, PyPDF2, python-docx, scikit-learn, spaCy, nltk, requests.

4. Resume Builder

  • Frontend: Vite/React candidate UI.
  • Backend: Python rendering/export pipeline.
  • Templates: LaTeX files in templates/ and Python PDF generation helpers.
  • Used to generate a polished resume/CV output from structured content.

5. Job Pipeline Creation

  • Uses Next.js API routes for pipeline creation, opportunity publishing, application tracking, and analytics.
  • Supabase stores pipelines, applications, resumes, interviews, and reports.
  • Zod validation and shared API helpers keep route handling consistent.

6. Video Interview

  • Flask app in ai-video-interviewer/ handles the interview experience.
  • Uses speech-to-text, text-to-speech, and a token-secured interview session flow.
  • FFmpeg is required for audio processing.

7. Reports Generated

  • Interview and resume analysis reports are generated as PDFs.
  • HR reports include evaluation summaries, KPI breakdowns, strengths, weaknesses, and transcript data.
  • Candidate-safe reports are also supported where appropriate.

LLMs and AI Services

The project uses Google Gemini in two main places:

  • gemini-2.0-flash in the Vite/FastAPI resume-analysis backend.
  • gemini-1.5-flash in the interview evaluator for structured interview assessment.

Other AI-related packages and services used in the repo:

  • @google/generative-ai
  • google-genai
  • speech_recognition
  • gTTS
  • gradio_client
  • pdf-lib
  • reportlab
  • fpdf2

Required Environment Variables

Keep real values in local .env files only. Do not commit secrets.

Root .env.local

NEXT_PUBLIC_SUPABASE_URL=...
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY=...
NEXT_PUBLIC_SUPABASE_ANON_KEY=...
SUPABASE_SERVICE_ROLE_KEY=...
NEXT_PUBLIC_APP_URL=http://localhost:3000
NEXT_PUBLIC_VIDEO_INTERVIEW_URL=http://localhost:5001
NEXT_PUBLIC_CLIENT_SIDE_BACKEND_URL=http://localhost:8000
NEXT_PUBLIC_CLIENT_SIDE_FRONTEND_URL=http://localhost:8080
GMAIL_USER=...
GMAIL_APP_PASSWORD=...
INTERVIEW_TOKEN_SECRET=...
GOOGLE_API_KEY=...
GEMINI_API_KEY=...

Client Side .env

GEMINI_API_KEY=...
VITE_BACKEND_URL=http://localhost:8000

Notes on keys

  • NEXT_PUBLIC_* values are required by the Next.js client/server routes.
  • SUPABASE_SERVICE_ROLE_KEY is used only on the server side.
  • GOOGLE_API_KEY / GEMINI_API_KEY power AI scoring and interview evaluation.
  • Never store real API keys in the README or commit them to git.

Quick Start

Prerequisites

  • Node.js 18+
  • Python 3.11+
  • FFmpeg for the video interview service
  • A Supabase project
  • Google Gemini API key

Install

git clone https://github.com/AaryanGole26/ELEVATR.git
cd ELEVATR
npm run install-all

Python setup

python -m venv venv
.\venv\Scripts\activate
pip install -r "Client Side/backend/requirements.txt"
pip install -r "ai-video-interviewer/requirements.txt"

Start services

start-all.bat

Or run them separately:

npm run dev
cd "Client Side/ai-skill-analyzer-main" && npm run dev
cd "Client Side/backend" && python main.py
cd ai-video-interviewer && python app.py

Ports

Service Port
Next.js portal 3000
Vite candidate app 8080
FastAPI backend 8000
Flask video interviewer 5001

Repository Layout

  • app/ - Next.js UI and API routes
  • ai/ - Shared AI helpers for interviewing, screening, and PDF reports
  • Client Side/ - Candidate resume analysis and resume builder experience
  • ai-video-interviewer/ - Flask-based video interview app
  • shared/ - Cross-app utilities, auth, Supabase helpers, and shared types
  • scripts/ - Utility scripts for storage setup and flow verification

Documentation

Additional context is available in:

  • BOOKLET.md
  • README_DOCUMENTATION_INDEX.md
  • START_HERE.md
  • QUICK_START.md
  • DEBUG_INTERVIEW_FLOW.md

Security Reminder

  • Never commit .env files or generated credentials.
  • Keep Supabase service role keys, Gemini keys, Gmail passwords, and interview secrets local only.
  • Generated reports and model artifacts should remain untracked.

Project Structure (Summary)

ELEVATR/
├── app/                      Next.js App Router
├── ai/                       Shared AI engine (Gemini)
├── shared/                   Auth, Supabase clients, email utils
├── scripts/                  One-off setup scripts
├── Client Side/
│   ├── ai-skill-analyzer-main/  Vite/React SPA
│   └── backend/                 FastAPI
├── ai-video-interviewer/     Flask interview app
├── BOOKLET.md                Full technical documentation
├── .gitignore
├── .env.local                (not committed — see above)
└── start-all.bat             Start all services

Tech Stack

  • AI: Google Gemini (generativeai), sentence-transformers, spaCy
  • Frontend: React 18, Vite, Next.js 14, Tailwind CSS, shadcn/ui, Radix UI
  • Backend: FastAPI (Python), Flask (Python)
  • Database: Supabase (PostgreSQL + Auth + Storage)
  • Email: Nodemailer (Gmail SMTP)
  • PDF: pdfplumber, PyMuPDF, WeasyPrint / pdfkit

License

MIT © 2026 Ayush Mayekar, Aaryan Gole, Dnyanesh Panchal, Vaibhav Narute, Tejashree Karekar, Charmi Jani

About

Unified AI-powered platform for resume analysis, CV building, video interviews, and HR pipeline management.

Resources

Readme
MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Contributors

Languages