All work

Data Engineering · Operational AI · Anomaly Detection

Intelligent Supply Chain Exception Control Tower

An operational control tower over a synthetic warehouse-and-transport event stream. It detects exceptions (late shipments, picking delays, SLA-breach risk, stockouts), scores delivery-delay risk, and ranks what needs attention now with a transparent priority score. It runs live on this page.

Domain
Supply chain / operations intelligence
Role
Data modeling, exception engine, risk scoring, product engineering
Stack
TypeScript · React · deterministic analytics · rules + risk scoring

Context

Operations teams often see what happened long after it matters. The harder question is what requires attention right now, across orders, warehouses, inventory, and shipments.

Problem

The goal was a control tower that surfaces and ranks live exceptions with enough evidence to act: which orders are late or about to breach SLA, which are stuck in the warehouse, and which SKUs are heading for stockout.

Architecture

Synthetic event streamOrders, warehouses, SKUs, stages and timings (seeded, reproducible)
Standardized event modelReceived → picking → packing → dispatch → transit → delivered
Feature & dwell calculationsStage dwell, SLA proximity, inventory cover
Exception engineLate shipments, picking and packing delays, SLA risk, stockouts, each with evidence
Delay-risk & priority scoreTransparent, decomposed factors, not a mystery score
Control towerLive overview, exception queue, root cause, inventory risk

Engineering

  • Seeded, reproducible synthetic operational event stream modeled after common warehouse/transport workflows (not a real ERP/WMS)
  • Standardized order-lifecycle event model with per-stage dwell timing
  • Exception engine detecting late shipments, picking/packing delays, SLA-breach risk, and stockout risk, each with cited evidence
  • Interpretable delivery-delay risk from SLA position, dwell, transit and priority
  • Transparent priority score decomposed into severity, value, customer priority, and risk
  • Live control-tower UI: overview, prioritized exception queue, root-cause breakdown, and inventory risk

Outcome

A working control tower you can use on this page: watch orders across stages, work a prioritized exception queue, open any exception for its evidence and priority breakdown, and see which SKUs are most exposed. It shows operational intelligence honestly, over a clearly-synthetic stream.

Live application

Open the control tower

Synthetic demonstration data · control tower over a simulated event stream

31Active orders
42Open exceptions
21Critical
€1.2MValue at risk
5Received
6Picking
2Packing
2Dispatched
16InTransit
3Delivered

Exception queue

Prioritized by severity, value, customer priority, and delay risk.

Root cause

Late shipment · Helios Group · ord-1031

Past promised 2d SLA (age 4d), still in transit via NordExpress.

Recommended: Escalate to carrier; notify customer with revised ETA.

Priority breakdown

Severity (Critical)
+30
Order value €60k
+15
Customer priority (Critical)
+24
Delay risk
+13

Inventory risk

SKUs ranked by stockout exposure (days of cover vs. lead time).

APP-1004Apparel1.9d cover · lead 11d
83
IND-1003Consumables2.8d cover · lead 13d
78
ELE-1013Consumables4.1d cover · lead 11d
63
CON-1001Apparel4d cover · lead 9d
56
CON-1002Home goods2d cover · lead 4d
50
ELE-1007Home goods6.3d cover · lead 12d
48
CON-1011Consumables9.4d cover · lead 14d
33
HOM-1010Apparel9.3d cover · lead 13d
28

Want to catch operational problems before they blow up?

Open the control tower above. Like it? Let's build one on your real operations.