CareerVerse - Career Development Ecosystem Powered by AI Agents

An AI-powered personalized career recommendation platform built using a monorepo architecture, consisting of Frontend (React/Vite), Backend (FastAPI), and AI-Core services (PhoBERT, vi-SBERT, NeuMF).

---

Team Information

Item	Information
Team Code	C2SE.17
Project Title	CareerVerse - Career Development Ecosystem Powered by AI Agents
Mentor	Nguyen Hai Minh

Team Members

No.	Full Name
1	Pham Tung Duong
2	Le Thanh Thien
3	Nguyen Cong Thinh
4	Tran Chi Tho
5	Pham Hoang Thuong

---

Table of Contents

1. Introduction
2. Core Features
3. System Architecture
4. Technology Stack
5. Project Structure
6. Subscription Plans
7. Installation Guide
8. Environment Configuration
9. Payment Testing Guide

---

Introduction

CareerVerse analyzes user personality traits through RIASEC and Big Five assessments, combined with essay analysis, to provide personalized career recommendations.

The platform integrates multiple AI techniques:

• Text analysis using PhoBERT and vi-SBERT (768-dimensional embeddings)
• Vector similarity search powered by pgvector
• Career ranking using NeuMF/MLP (Neural Matrix Factorization)
• Real-time recommendation optimization using Thompson Sampling
• AI-powered career assistant and mock interviews through Google Gemini API

---

Core Features

AI Career Recommendation

• Combines four signals:

  • Embedding Similarity (20%)
  • RIASEC Matching (45%)
  • Big Five Matching (10%)
  • NeuMF Prediction (25%)

• Thompson Sampling continuously learns from user feedback

• Database containing over 900 occupations based on the O*NET standard

AI Mock Interview

Text Mode

• Interview questions generated by Gemini AI
• Multi-dimensional answer evaluation
• Detailed strengths, weaknesses, and improvement suggestions

Voice Mode

• Real-time Speech-to-Text and Text-to-Speech
• Powered by Faster-Whisper and Edge TTS

Evaluation Criteria

• Technical Knowledge
• Communication Skills
• Logical Thinking
• Experience
• Professional Attitude

Skill Gap Analysis

• Upload CVs in PDF, DOCX, JPG, or PNG format

• Automatic skill extraction

• Compare current skills against target career requirements

• Categorize skill gaps into:

  • Critical
  • Important
  • Nice-to-have

• Recommend courses from Coursera, Udemy, and edX

Personalized Learning Roadmap

• Generate learning paths based on identified skill gaps
• Progress tracking
• Email reminders
• Integration with a catalog of over 1,000 courses

Mentor Matching

• Semantic similarity-based mentor matching
• RIASEC compatibility scoring
• Real-time chat using WebSocket

Labor Market Intelligence

• Automatic crawling from VietnamWorks, ITViec, and TopCV every six hours
• Industry demand trend analysis
• Salary statistics by experience level

---

System Architecture

Frontend (React + Vite)
        │
        │ HTTP / WebSocket
        ▼
Backend (FastAPI - Port 8000)
        │
        ├── PostgreSQL + pgvector   ← Primary storage & vector search
        ├── Neo4j                   ← Career-Skill Knowledge Graph
        ├── Redis                   ← Session caching
        │
        │ Internal HTTP Call
        ▼
AI-Core Service (FastAPI - Port 9000)
        │
        ├── PhoBERT / vi-SBERT      ← NLP & Embeddings
        ├── NeuMF MLP               ← Collaborative Filtering
        └── Thompson Sampling       ← Bandit Optimization

---

Technology Stack

Frontend

• React 18 + TypeScript — UI Framework
• Vite — Build Tool
• Tailwind CSS — Styling
• React Router — Routing
• React Query — Data Fetching & Caching
• Axios — HTTP Client

Backend

• FastAPI — Web Framework
• SQLAlchemy + psycopg — ORM & PostgreSQL Driver
• Pydantic — Data Validation
• APScheduler — Background Job Scheduling
• Redis — Caching
• JWT — Authentication

Database

• PostgreSQL 14+ with pgvector — 768-dimensional Vector Storage
• Neo4j 5 — Career-Skill Relationship Graph
• Redis — Session & Runtime Cache

AI / Machine Learning

• PyTorch — Deep Learning (NeuMF)
• HuggingFace Transformers — PhoBERT, vi-SBERT
• FAISS — Vector Similarity Search
• scikit-learn — Classical Machine Learning Algorithms
• Faster-Whisper — Real-Time Speech-to-Text
• Edge TTS — Natural Text-to-Speech

External APIs

• Google Gemini API — Chatbot, Interview Question Generation, Answer Analysis
• Cloudflare R2 — CV File Storage
• VNPay / ZaloPay — Payment Gateways

Infrastructure

• Docker & Docker Compose — Containerization
• Uvicorn — ASGI Server

---

Project Structure

AI-Based-Career-Recommendation-System/
├── apps/
│   ├── backend/
│   └── frontend/
├── packages/
│   └── ai-core/
├── db/
├── docker-compose.yml
└── README.md

Backend Modules

• Authentication
• User Management
• Assessments
• Career Catalog
• AI Recommendation Engine
• Skill Gap Analysis
• Mock Interview
• Learning Path
• Mentor Matching
• AI Chatbot
• Real-time Chat
• Course Catalog
• Job Market Data
• Payments
• Subscription Management
• PDF Reports
• Analytics
• Content Management
• Market Trends

---

Subscription Plans

Feature	Free	Basic (99,000 VND)	Premium (199,000 VND)	Pro (299,000 VND)
Assessments per Month	5	20	Unlimited	Unlimited
Career Recommendations	1	5/month	Full Access	Full Access
Learning Roadmaps	Level 1	Levels 1–2	Full Access	Full Access
AI Chatbot	—	—	—	24/7
Voice Interview	—	—	—	✓
PDF Export	—	—	—	✓
RIASEC + Big Five Analysis	—	✓	✓	✓

---

Installation Guide

System Requirements

Software	Version	Verification
Docker Desktop	Latest	docker --version
Python	3.11+	python --version
Node.js	18+	node --version

Minimum Requirements

• 8GB RAM (16GB Recommended)
• 10GB Free Storage

---

Step 1 — Start Databases

cd AI-Based-Career-Recommendation-System

docker compose --env-file apps/backend/.env up -d

docker compose ps

Import Database Snapshot

docker compose exec -T postgres psql -U postgres -d postgres -c "SELECT pg_terminate_backend(pid) FROM pg_stat_activity WHERE datname='career_ai';"

docker compose exec -T postgres psql -U postgres -d postgres -c "DROP DATABASE IF EXISTS career_ai;"

docker compose exec -T postgres psql -U postgres -d postgres -c "CREATE DATABASE career_ai;"

docker compose cp db/backup/dev_snapshot_utf8.sql postgres:/tmp/dev_snapshot_utf8.sql

docker compose exec -T postgres psql -U postgres -d career_ai -v ON_ERROR_STOP=1 -f /tmp/dev_snapshot_utf8.sql

docker compose exec redis redis-cli ping

curl http://localhost:7474

---

Step 2 — Run AI-Core Service (Port 9000)

cd packages/ai-core

python -m venv .venv

.venv\Scripts\activate

pip install -r requirements.txt
pip install -e .

uvicorn src.api.main:app --reload --port 9000

---

Step 3 — Run Backend Service (Port 8000)

cd apps/backend

python -m venv .venv

.venv\Scripts\activate

pip install -r requirements.txt

pip install -e ../../packages/ai-core

pip install sentence-transformers>=2.7.0 neo4j>=5.0.0

pip install beautifulsoup4 lxml fake-useragent

pip install PyPDF2==3.0.1 edge-tts

uvicorn app.main:app --reload --port 8000

---

Step 4 — Run Frontend (Port 5173)

cd apps/frontend

npm install

npm run dev

---

Access URLs

Service	URL
Main Application	http://localhost:5173
Backend API Docs	http://localhost:8000/docs
AI-Core API Docs	http://localhost:9000/docs
Neo4j Browser	http://localhost:7474
PostgreSQL	localhost:5433

---

Environment Configuration

Create a .env file inside apps/backend/:

DATABASE_URL=postgresql://postgres:123456@localhost:5433/career_ai
AI_CORE_BASE=http://localhost:9000
ALLOWED_ORIGINS=http://localhost:5173
SECRET_KEY=your_secret_key

VNPAY_TMN_CODE=your_vnpay_tmn_code
VNPAY_HASH_SECRET=your_vnpay_hash_secret

ZALOPAY_APP_ID=your_zalopay_app_id
ZALOPAY_KEY1=your_zalopay_key1
ZALOPAY_KEY2=your_zalopay_key2

GEMINI_API_KEY=your_gemini_api_key

---

Payment Testing Guide

The platform integrates VNPay and ZaloPay in Sandbox mode.

VNPay Sandbox

Field	Value
Bank	NCB
Card Number	9704198526191432198
Cardholder Name	NGUYEN VAN A
Issue Date	07/15
OTP	123456

ZaloPay Sandbox

Field	Value
Card Number	4111111111111111
Cardholder Name	NGUYEN VAN A
Expiry Date	06/26
CVV	123

---

Contact

For questions or support, please contact:

Email: tranchitho160704@gmail.com