Slope-Sensor Data AI Analysis Platform

Context

An industry-academia collaboration between a Curtin University research laboratory and an industry partner, focused on extracting operational insight from real-world geotechnical slope-monitoring data (tilt sensors, time-series readings from multiple sites).

The research team needed a tool that non-specialists — engineers and operators, not ML researchers — could use to spot patterns and produce decision-grade reports.

Challenge

• Raw geospatial / slope sensor data is dense, multi-dimensional, and time-series heavy
• Matching patterns across sites and periods required dynamic-time-warping-class techniques, not naïve comparison
• The audience for the output was non-technical — reports had to be written in plain language, not plots alone
• Research-grade notebooks weren't enough; the partner needed a deployable web tool

Solution

A full-stack web application that ingests slope-monitoring data and produces automated analytical reports:

• Backend (Python / Flask on AWS): ingestion, FastDTW-based time-series similarity, data cleaning pipelines
• Frontend visualisation: D3.js + AmCharts interactive views — multiple chart types for comparison across sites and time windows
• AI report layer: OpenAI API wraps the numeric analysis in natural-language summaries, automatically drafting analytical reports that a non-technical reader can consume
• Report output: Word/DOCX template pre-populated with generated findings, ready for distribution

Results

• Delivered to the research lab and handed to the industry partner
• Non-specialist users can run analyses and get readable reports without touching a notebook
• Demonstrated that AI can augment — not replace — specialist judgement in geo-engineering contexts

What this proves

SJSoftware handles projects beyond commercial SaaS — research collaborations, science-facing visualisation, and AI-augmented analytical tooling are in scope.