Built a real-time operations dashboard replacing 14 Excel reports, giving floor managers a single source of truth for production KPIs.
Floor managers at a 3-facility manufacturing company started each shift by manually compiling 14 Excel reports from 6 systems — a 3-hour process per manager per day. Data was always stale, KPIs were inconsistent across facilities, and no one had a real-time view of production throughput.
A single-page React dashboard integrating data from 6 source systems via a Node.js aggregation API. D3.js powers live production line charts, OEE gauges, and downtime Pareto charts. WebSockets push updates every 30 seconds. LDAP integration with Active Directory for zero-friction SSO for all 200+ staff.
3 hours of daily report-building eliminated per manager (6 managers = 18 hours/day across the company). All 14 KPI definitions standardized. Floor managers can now react to downtime events in under 2 minutes instead of the next-shift briefing.
SYSTEM ARCHITECTURE
Industrial IoT operations architecture interfacing SCADA shopfloor hardware and legacy ERPs with high-frequency WebSocket distribution, normalized PostgreSQL time-series, and D3.js visualization.
Isolated fragile legacy plant hardware (Modbus/OPC-UA) behind containerized protocol adapters, allowing the web interface to iterate without touching factory machine controllers.
Streamed compressed raw timeseries deltas to client browsers where D3.js computed Pareto charts dynamically, reducing central server CPU overhead by 78%.
TECHNICAL DEPTH
The 6 source systems include a 20-year-old SCADA platform, 2 ERP instances, and 3 internal databases. We built adapter modules per source with a normalized data model in PostgreSQL. ETL runs every 30 seconds with conflict resolution for overlapping data.
D3.js for production throughput sparklines, OEE donut charts, and Pareto downtime charts. Recharts for the simpler KPI trend lines. All charts are interactive — drill-down from facility → line → machine. Export to PDF/CSV for shift handover reports.
LDAP sync with Active Directory assigns roles based on AD groups. FLOOR_MANAGER sees one facility. PLANT_DIRECTOR sees all 3. ENGINEER sees maintenance-specific views. Row-level security in PostgreSQL enforces data scoping — the UI permissions are decorative only.