Open-source federal workforce intelligence. Browse, explore, and visualize U.S. federal employee compensation and workforce data from the Office of Personnel Management (OPM).
 |
 |
 |
 |
LevelsGov makes OPM FedScope data accessible and understandable. Instead of downloading bulk CSV files from OPM, you get an interactive dashboard with search, filters, and visualizations covering 2M+ federal employee records across every agency, occupation, and state.
Think of it as levels.fyi but for the federal government.
- Key workforce stats: total employees, median pay, agency count, new hires, net change
- Interactive choropleth map of federal pay by state
- Top paying agencies and occupations
- Pay breakdowns by tenure, education, age, and STEM classification
- GS grade distribution, work schedule breakdown, supervisory pay gap
- Workforce trend analysis: separation reasons, agency changes, STEM brain drain
- Search and filter the full federal workforce dataset
- Filter by agency, state, occupation, grade, pay plan, education, age, work schedule, and more
- Sortable, paginated results
- Browse accession and separation records with the same filter capabilities
- Understand who's joining and leaving federal service, and why
- Interactive treemap of 500+ federal occupations sized by headcount and colored by AI exposure score
- Demographic breakdowns (age, education) for each occupation
- Custom scoring methodology based on digital work, routine cognition, interpersonal labor, and regulatory criticality
- Hierarchical view of the federal government structure
- Cabinet departments, military branches, independent agencies, legislative and judicial branches
- Agency subelements with headcount data
- Framework: Next.js 16 (React 19, TypeScript)
- Database: PostgreSQL
- Styling: Tailwind CSS v4, shadcn/ui
- Charts: Recharts, react-simple-maps
- Deployment: Docker, Coolify (self-hosted)
- Node.js 20+
- PostgreSQL 14+
- Python 3.10+ (for the data pipeline)
git clone https://github.com/mishafyi/levelsgov.git
cd levelsgov
npm installThe app connects to PostgreSQL via DATABASE_URL (default: postgresql://localhost:5433/fedwork). Create the database and load the schema:
createdb fedwork
psql postgresql://localhost:5433/fedwork -f scripts/schema.sqlCreate a .env.local in the project root:
# Postgres connection (omit to use the default above)
DATABASE_URL=postgresql://localhost:5433/fedwork
# Optional: token guarding the on-demand cache-revalidation route (/api/revalidate)
# REVALIDATE_TOKEN=your-secretThe Python pipeline downloads OPM FedScope files and bulk-loads them into Postgres:
pip install psycopg2-binary requests
python3 scripts/download.py --months 18 # download + import ~18 monthsUseful flags: --dataset employment (single dataset), --dry-run (preview only), --no-import (download without importing). To load one file directly, use python3 scripts/import.py <dataset> <file>.
npm run devOpen http://localhost:3000. A Dockerfile is included for containerized/self-hosted deployment.
All workforce data comes from the OPM FedScope public dataset. AI exposure scores are generated using a custom methodology documented in the repository.