Tech Job Recommender

A smart job matching system for software engineers, data scientists, and DevOps folks. Enter your skills and preferences, get personalized job recommendations with clear explanations of why each job fits you.

🚀 Live Demo

Component	URL
Frontend (Dashboard)	jobmatch-tech.streamlit.app
Backend (API)	jobmatch-production-0420.up.railway.app
API Docs	jobmatch-production-0420.up.railway.app/docs

What It Does

• Matches your skills against 2,600+ real tech jobs from LinkedIn
• Shows you exactly why each job is a good (or not so good) fit
• Lets you adjust what matters most: skills, salary, remote work, experience level
• Gives you skill gap analysis so you know what to learn next

Tech Stack

• Frontend: Streamlit, Plotly
• Backend: FastAPI, SQLite
• ML: Sentence-Transformers, Qdrant
• LLM: Groq (Llama 3)

Quick Start

1. Clone and Install

git clone https://github.com/your-repo/job-recommender.git
cd job-recommender
pip install -r requirements.txt

2. Set Up Your API Key

You need a free Groq API key for the LLM features. Get one at console.groq.com.

cp .env.example .env
# Edit .env and add your GROQ_API_KEY

3. Download Job Data

# Set up Kaggle credentials first (see below)
python -m scripts.download_kaggle_data --dataset linkedin
python -m scripts.ingest_data --kaggle data/kaggle/linkedin/postings.csv --limit 50000

4. Generate Embeddings

python -m scripts.generate_embeddings
python -m scripts.index_embeddings

5. Run the App

# Start the dashboard
streamlit run streamlit_app/app.py

Open http://localhost:8501 in your browser.

How It Works

Your Profile → Ranking Engine → Recommendations
     ↓              ↓                  ↓
  Skills      Weighted Scoring    Explanations
  Experience  Skill Matching      Skill Gaps
  Salary      Location Match      Match Scores

The system uses:

• Sentence Transformers for understanding job descriptions
• Qdrant for fast similarity search
• Groq/Llama for extracting requirements from job posts
• Weighted scoring across 5 dimensions: skills, experience, seniority, location, salary

Weight Presets

Pick a preset that matches your job search priorities:

Preset	Focus
Skill-Focused	Best skill match (50% weight on skills)
Career Growth	Experience alignment (35% on experience)
Compensation First	Salary fit (35% on salary)
Remote Priority	Location/remote (30% on location)
Balanced	Equal consideration of everything

Or set your own custom weights.

Project Structure

├── src/                    # Core logic
│   ├── ranking.py          # Job scoring and ranking
│   ├── recommendation.py   # Explanation generation
│   ├── embedding.py        # Text embeddings
│   ├── database.py         # SQLite storage
│   └── vector_store.py     # Qdrant integration
├── streamlit_app/          # Web dashboard
├── scripts/                # Data pipeline
└── tests/                  # Unit and property tests

API Endpoints

If you want to use the REST API directly:

# Start the API server
uvicorn src.api.main:app --port 8000

Endpoint	Description
POST /api/candidates	Create a candidate profile
GET /api/recommendations/{id}	Get job recommendations
GET /api/jobs	List all jobs
PUT /api/weights/{id}	Update ranking weights
POST /api/feedback	Submit like/dislike feedback

API docs at http://localhost:8000/docs

Setting Up Kaggle

To download the LinkedIn job dataset:

1. Create account at kaggle.com
2. Go to Settings → API → Create New Token
3. Save kaggle.json to ~/.kaggle/ (Linux/Mac) or %USERPROFILE%\.kaggle\ (Windows)
4. Run python -m scripts.download_kaggle_data --check

Tech Stack

• Frontend: Streamlit, Plotly
• Backend: FastAPI, SQLite
• ML: Sentence-Transformers, Qdrant
• LLM: Groq (Llama 3)

Deployment

This project uses a split deployment:

• Backend (API): Railway - jobmatch-production-0420.up.railway.app
• Frontend (Dashboard): Streamlit Cloud - jobmatch-tech.streamlit.app

Step 1: Deploy Backend to Railway

1. Push your code to GitHub

2. Go to railway.app and sign in with GitHub

3. Click "New Project" → "Deploy from GitHub repo"

4. Select your repository

5. Add environment variable: GROQ_API_KEY

6. Railway will build using the Dockerfile and deploy

7. Copy your Railway URL (e.g., https://your-app.railway.app)

Step 2: Deploy Frontend to Streamlit Cloud

1. Go to share.streamlit.io

2. Sign in with GitHub

3. Click "New app" and select your repository

4. Set these options:

   • Main file path: streamlit_app/app.py
   • Python version: 3.11

5. Add secret in "Advanced settings" → "Secrets":

   API_BASE_URL = "https://your-app.railway.app"

6. Click "Deploy"

Your frontend will be live at https://your-app.streamlit.app