frapax/peoplemesh repository overview

⁠PeopleMesh

The right match in your mesh.

PeopleMesh is the AI-powered matching layer for modern organizations. It helps teams discover the right colleagues, internal opportunities, communities, and projects through semantic search that understands context, not just keywords.

By combining embeddings with metadata-based ranking, PeopleMesh surfaces high-signal matches faster, improves internal mobility, and reduces discovery noise.

Privacy and trust are first-class: PeopleMesh includes granular consent controls, GDPR-aligned rights handling, configurable retention, and pseudonymized audit trails.

⁠Why this image

This image packages PeopleMesh as a production-style Quarkus service with:

• REST API endpoints for app and integrations
• MCP server support for AI assistants and agents
• PostgreSQL + pgvector compatibility for semantic search storage
• Ollama-based default model setup (granite4:3b + granite-embedding:30m)
• Configurable provider flows for alternative AI backends

⁠Quick start

docker pull frapax/peoplemesh:latest

Start the full stack (PeopleMesh + dependencies):

docker compose -f tools/compose/docker-compose.dependencies.yml up -d

Where to get this compose file:

• In the repository at tools/compose/docker-compose.dependencies.yml
• On GitHub: https://github.com/francescopace/peoplemesh/blob/main/tools/compose/docker-compose.dependencies.yml⁠

Set environment variables before startup (secrets, OIDC credentials, CORS, optional image overrides).

After first startup, pull the default Ollama models:

docker exec -it peoplemesh-ollama ollama pull granite4:3b
docker exec -it peoplemesh-ollama ollama pull granite-embedding:30m

⁠Runtime dependencies

PeopleMesh depends on these external services at runtime:

• PostgreSQL with pgvector extension enabled (required)
  Used for relational data and vector similarity search.
• Ollama (required by default setup)
  Used for chat + embedding models (granite4:3b, granite-embedding:30m by default).
• Docling service (required if CV import is enabled)
  Used to parse CV documents for profile enrichment.
• OIDC identity provider (required)
  At least one supported provider must be configured (Google or Microsoft) for login flows.

If you use non-default AI providers, see the configuration reference for supported alternatives and related environment variables.

⁠Full environment variable reference

For the complete and up-to-date list of available variables, see:

• https://github.com/francescopace/peoplemesh/blob/main/docs/reference/configuration.md⁠
• Compose guide: https://github.com/francescopace/peoplemesh/blob/main/docs/how-to/run-compose-stack.md⁠

⁠Verify runtime

• App: http://localhost:8080
• Health: GET /q/health
• Bootstrap info: GET /api/v1/info

⁠Documentation

• Main repository: https://github.com/francescopace/peoplemesh⁠
• Quick start: https://github.com/francescopace/peoplemesh/blob/main/docs/getting-started/quickstart.md⁠
• Container run guide: https://github.com/francescopace/peoplemesh/blob/main/docs/how-to/build-and-run-container.md⁠
• Docker Compose stack guide: https://github.com/francescopace/peoplemesh/blob/main/docs/how-to/run-compose-stack.md⁠
• Configuration reference: https://github.com/francescopace/peoplemesh/blob/main/docs/reference/configuration.md⁠
• API reference: https://github.com/francescopace/peoplemesh/blob/main/docs/reference/api.md⁠
• Support options: https://github.com/francescopace/peoplemesh/blob/main/SUPPORT.md⁠

⁠License

Apache License 2.0.