Simon Stirling.Download CVContact
All projects
04 / INDEPENDENT ENGINEERING DEMONSTRATORPublic without sign-in
Data quality & lineage

Traceability LabEvery corrected reading keeps its original story.

Ingest synthetic manufacturing measurements, investigate known quality defects and follow a reviewed correction from its original source to its current overlay.

Open live demoView repository
Traceability Lab showing synthetic manufacturing data quality and source traceability.
The public application with synthetic example data.
01 / FUNCTION & ABILITIES

Every corrected reading keeps its original story.

01

Reconcile the ingestion

Retain raw records and reconcile usable, quarantined, duplicate and missing expected measurements.

02

Investigate known defects

Check completeness, exact and conflicting duplicates, types, units, ranges, timestamps, equipment and sampling-plan membership.

03

Review a bounded correction

Operators propose permitted field changes. Reviewer decisions check versions and findings before applying a curated overlay.

04

Follow the evidence

Inspect original/current fields, source pointers, resolved finding history, a hash-linked audit and a verifiable six-file evidence ZIP.

02 / PLANNED BOUNDARIES, VISIBLE FLOW

From the whiteboard
to a working application.

Play a scenario, step through the decisions or select a component to inspect its responsibility. The diagrams follow the implemented design.

THE ARCHITECTURE BOARD04 / Traceability Lab
Illustrated flow
Manufacturing data quality & traceability demonstrator: architecture and data flowThe arrows illustrate ingestion and review steps. Demo roles belong to one visitor; approved corrections change a curated overlay while the raw source remains intact. Select a component or use the readable step list below.SYNTHETIC INPUTSHARED POLICY SERVICE + SQLITE / D1 + INVESTIGATION UIsource rowsretain + checklineageknown defectsproposed repairexplicit reviewapproved overlay + auditThe sample deliberately includes missing slots, duplicate readings, malformed values and inconsistent units.Synthetic source filesBatches + measurementsThe service retains source records and evaluates completeness, identity, types, units, ranges, time and equipment.Ingestion + checksValidate and reconcileRaw source rows and hashes are immutable. Curated measurements and finding history are versioned in SQLite or D1.Retained SQL recordsRaw + curated overlayThe investigation connects original and current fields to imports, source rows, proposals, decisions and the audit chain.Lineage + evidencePointers / audit / ZIPFindings explain why readings are quarantined or expected slots are missing. Resolved findings remain inspectable.Quality findingsOpen + resolved historyA permitted value, exact unit, timestamp or equipment correction is validated as a proposal. Identity changes are outside this workflow.Operator proposalNo immediate changeReviewer scope, proposal state, measurement version and finding improvement are checked before the controlled transaction.Reviewer decisionReason + version checkfollow the evidence, then the decision
Step 1 / 4 · click any component
FOLLOW THE FLOW

Admit the sample files

Synthetic batch masters, the sampling plan and measurement records enter the ingestion service.

Read all 4 steps without animation

See which source rows can be used and why the others cannot.

The arrows illustrate ingestion and review steps. Demo roles belong to one visitor; approved corrections change a curated overlay while the raw source remains intact. Arrows and moving dots are an explanation of the implementation, not live telemetry.
03 / A FIVE-MINUTE WALKTHROUGH

Try it for yourself.

Keep the live demo open
  1. 01

    Ingest the sample run

    Use the Operator role. Eleven source rows reconcile to four usable, five quarantined and two duplicates; eight of nine expected slots are observed.

  2. 02

    Trace a finding

    Open the UNIT finding for MES-004 in batch AX-1102. Follow the source-row pointer, inspect raw JSON and compare original/current fields.

  3. 03

    Propose the correction

    Use the fictional source-owner clarification to change degC to °C. Submission creates a proposal and leaves the measurement unchanged.

  4. 04

    Review and approve

    Switch to Reviewer, record a rationale and approve. The curated measurement becomes version 2 while the original and resolved finding remain visible.

  5. 05

    Verify the trail

    Open Audit trail and download the six-file ZIP. Optionally ingest the separate follow-up pressure reading to close AX-1103's completeness gap.

04 / ENGINEERING JUDGEMENT

Why it works this way.

Retain the original

Corrections update a versioned overlay. Canonical source hashes, row ordinals and the original reading remain available for investigation.

Protect concurrent reviews

A mutation token and optimistic revision guard the SQL transaction. Approval also checks the proposed measurement version.

Verify before export

Export verification checks source hashes, reconciliation, audit links and proposal/decision history before producing the evidence files.

05 / INSPECT THE EVIDENCE

Tested behaviour.
Explicit limits.

  • 40 unit tests and the same 24 HTTP scenarios on local SQLite, Worker/D1 and both public origins are recorded.

  • Local restart and public redeployment checks preserve source records, reviewed overlays, finding history and access.

  • Published browser validation includes actual downloads, independent verification, re-import and mobile layout.

Read the recorded validation