Skip to content

SUBMISSION_CHECKLIST.md

Latest commit

 

History

History
84 lines (66 loc) · 2.86 KB

SUBMISSION_CHECKLIST.md

File metadata and controls

84 lines (66 loc) · 2.86 KB

Submission Checklist

Implemented

  • Express + TypeScript backend with modular routes, controllers, services, validators, and shared MongoDB connection.
  • Mongoose models for Worker, Workstation, and AIEvent.
  • Seed data for 6 workers, 6 workstations, and 101 realistic shift events.
  • AI event ingestion at POST /api/events/ingest for single events and batches.
  • Zod validation for event payloads and metric query parameters.
  • Duplicate protection using deterministic event fingerprints.
  • Event listing at GET /api/events with filters.
  • Worker, workstation, and factory metrics endpoints.
  • Timestamp-based duration calculations that tolerate out-of-order arrival.
  • Admin seed/reset/clear utilities.
  • Swagger/OpenAPI docs at /docs/.
  • React dashboard at /, connected to live backend APIs.
  • Bootstrap endpoint at GET /api/bootstrap.
  • Dockerfile and Docker Compose setup with persistent MongoDB volume.
  • Healthcheck endpoint and Docker healthcheck.
  • Vitest/Supertest coverage for ingestion, validation, metrics, shell, and bootstrap behavior.

How To Run

Non-Docker

cp .env.example .env
npm install
npm run frontend:install
npm run seed
npm run build
npm run dev

Requires MongoDB to be running at the configured MONGODB_URI in .env. For local MongoDB, use mongodb://localhost:27017/factory_productivity.

Open:

Docker

docker compose up --build

Reset the Docker MongoDB volume:

docker compose down -v
docker compose up --build

Self-Check Commands

curl http://localhost:3000/health
curl -X POST http://localhost:3000/api/admin/seed
curl http://localhost:3000/api/bootstrap
curl http://localhost:3000/api/metrics/factory
curl http://localhost:3000/api/metrics/workers
curl http://localhost:3000/api/metrics/workstations

Assumptions

  • working, idle, and absent are state transitions.
  • A state starts at its timestamp and lasts until the next state event for the same worker/workstation pair.
  • Metrics sort by event timestamp, not insertion order.
  • product_count events are production increments and do not define active time.
  • Final open state intervals close at to when provided, otherwise at the max timestamp in the filtered dataset.
  • Duplicate retries are detected using a deterministic fingerprint from timestamp, worker, workstation, event type, and count.
  • Seed data is intentionally deterministic so evaluator checks are repeatable.

Intentionally Simplified

  • No authentication or authorization.
  • No queue or streaming ingestion layer.
  • No camera/site/model-version tables yet.
  • No background aggregation jobs.
  • Frontend is focused on the core productivity dashboard and keeps advanced drilldowns intentionally light.
  • MongoDB is used through Mongoose, with Docker Compose providing a local database.