SELECTED WORK / ML · APIS · ENGINEERING

MAINTENANCE SIGNAL

A prototype ML workflow focused on prioritisation, evaluation, false positives and human review rather than prediction alone.

← BACK TO WORK
THE QUESTION What should the system actually help a person decide or understand?
DATA FLOW Source inputs → validation → transformation → model / logic → review
ARCHITECTURE APIs, Python services, PostgreSQL and cloud components
EVALUATION Quality checks, failure cases, observability and human feedback
TRADE-OFFS Complexity, latency, false positives, maintainability and scope

Concept architecture, not a claim of measured customer impact.