Atlas Operations Platform
A single operating surface for logistics teams running high-volume, multi-site fulfilment.
Operations were coordinated across a dozen spreadsheets and three legacy tools that never agreed with each other. Every shift started with an hour of manual reconciliation, and exceptions were only discovered after they became customer problems.
We modelled the real dispatch workflow with the operations leads, then built a live planning board on top of an event-sourced core. Every status change flows through a single pipeline, so the board, the reports and the alerts are always looking at the same truth.
Live planning board
Drag-and-drop assignment with capacity, SLA and cost surfaced inline as decisions are made.
Exception routing
Rules engine detects at-risk jobs early and routes them to the right owner with full context.
Operational reporting
Shift, site and network views built from the same event stream that powers the board.
Audit timeline
Every job carries a complete, queryable history — who changed what, when and why.
- 01Frontend
Next.js app router, streaming server components, real-time board over WebSockets
- 02API Layer
Typed RPC gateway, authz at the edge, request-level tracing
- 03Services
Planning, dispatch and reporting services coordinated by Temporal workflows
- 04Database
Event-sourced Postgres core with read models materialised per view
- 05External Integrations
Carrier APIs, telematics feeds and finance system sync
Frontend
- Next.js
- TypeScript
- Tailwind CSS
- TanStack Query
Backend
- Node.js
- Temporal
- gRPC
Database
- PostgreSQL
- Redis
Cloud
- AWS ECS
- RDS
- CloudFront
DevOps
- Terraform
- GitHub Actions
- OpenTelemetry
- Shift planning consolidated into one system
- Exceptions surfaced before they reach customers
- Reporting drawn from a single source of truth
- Architecture that scales with new sites