Applied AI infrastructure · Operations
Manufacturing Event Investigation
An investigation workspace that reconstructs production history, surfaces conflicting events, and connects job reviews to their original records.
From raw log to reviewable evidence
The problem
A raw event log mixes production state, machine readings, duplicate messages, and conflicting revisions. Reviewers need to reconstruct what was known at a chosen time and trace the evidence behind each exception.
Implementation
Normalization retains conflicting variants and source references. Replay rebuilds job state at the selected cursor. The UI joins machine telemetry, searchable events, job comparisons, and role-scoped review workflows with database-backed notes and optimistic concurrency checks.
Evidence
The project README documents a synthetic log with 19,519 rows across 312 jobs. The local version includes tests for replay boundaries, Pacific-time windows, authorization, assignments, and concurrent review updates. The linked public source is a separate repository.
Limits
Waiting time and descriptive review signals do not establish downtime, utilization, financial savings, or a causal diagnosis. The public deployment requires sign-in. Current local collaboration changes may be newer than the public repository.