A browser-based salary prediction tool for Finnish tech professionals, powered by machine learning. Built using data from the Koodiklinikka Salary Survey 2024.
- Browser-based ML inference - No server required, predictions run entirely in your browser
- ONNX Runtime Web - Fast inference using WebAssembly
- Privacy-first - Your data never leaves your device
- Responsive design - Works on desktop and mobile
- React + TypeScript - UI framework
- Vite - Build tool
- styled-components - CSS-in-JS styling
- ONNX Runtime Web - Browser-based ML inference
- Random Forest Regressor - Model architecture
- scikit-learn - Model training
- skl2onnx - ONNX conversion
- Model Performance: R² = 0.44, MAE = €834
- Python - Data cleaning and analysis
- pandas - Data manipulation
- Jupyter Notebooks - Interactive analysis
| Notebook | Description | Kaggle |
|---|---|---|
| Data Download | Downloads raw survey data in Finnish and English | View |
| Data Cleaning | Cleans and processes the raw data | View |
| Data Analytics | Exploratory analysis and model training | View |
├── client/ # React application
│ ├── src/
│ │ ├── components/ # React components
│ │ ├── lib/ # ML model utilities
│ │ └── styles/ # Global styles
│ └── public/
│ ├── salary_predictor_v1.0.0.onnx # ML model
│ └── salary_predictor_v1.0.0_metadata.json
├── data/ # Data processing
│ ├── raw/ # Raw downloaded data
│ ├── processed/ # Cleaned data
│ ├── models/ # Trained models
│ └── *.ipynb # Jupyter notebooks
└── .github/workflows/ # CI/CD deployment
cd client
npm install
npm run devThe app will be available at http://localhost:5173
cd data
uv sync # Install dependenciesRun the notebooks in order:
finnish-tech-salary-survey-2024-data-download.ipynbfinnish-tech-salary-survey-2024-data-cleaning.ipynbfinnish-tech-salary-survey-2024-data-analytics.ipynb
The salary predictor uses a Random Forest model trained on 532 survey responses. Features include:
- Experience level
- Role
- Location
- Education level
- Company size
- Employment type
- Age group
This is an experimental prediction tool based on self-reported survey data from 532 respondents. The model explains approximately 44% of salary variance (R² = 0.44). Actual salaries can vary significantly based on factors not captured in this survey. Use predictions as rough estimates only, not for salary negotiations or financial decisions.
Survey data from Koodiklinikka Salary Survey 2024 - a Finnish developer community survey collecting salary information from tech professionals working in Finland.
MIT