The Nucleus
Replaces manual, paper- and spreadsheet-based inspection workflows for industrial Non-Destructive Testing (NDT) service providers with one auditable platform spanning field data capture, review, reporting, and client delivery.
Overview
A production inspection-management platform built for industrial NDT companies — teams running ultrasonic, radiographic, and visual inspections on pressure vessels, pipelines, and storage tanks. It covers the full inspection lifecycle across six role-specific portals (admin, manager, reviewer, inspector, client, IT), with a web app, a Windows desktop app, and an Android tablet app so inspectors can capture data in the field, online or offline, while everyone else works from a single source of truth.
Key Features
- Offline-first field data capture on desktop and Android tablet apps, with encrypted local storage and automatic background sync back to a central server
- AI-assisted inspection report generation, gated behind data-completeness checks and automated fact-checking so generated narrative can never contradict the recorded data
- Interactive 3D asset visualization with real-time thickness/corrosion heatmaps and automated asset health scoring
- Full role-based workflow: job assignment, review, correction requests, approval, and client sign-off, with a tamper-evident audit log across every action
- White-labelled, multi-deployment architecture so the same platform can be branded and rolled out to different clients independently
Architecture
React web portal + FastAPI/PostgreSQL backend, with Electron desktop and Android (Expo) field apps syncing offline-captured inspection data to a central server, plus an async pipeline for AI-assisted report generation
What I Learned
“This was the first system I built where the happy path was the easy part — most of the real engineering went into what happens when a tablet drops offline mid-inspection, two people edit the same record, or an AI-generated sentence risks contradicting the actual test data. Building for a regulated, safety-adjacent domain forced me to treat data integrity and failure handling as core features from day one, not edge cases bolted on later.”