Selected engineering challenges

Four problems that show the range of the work. Client names stay private, and no specific performance figure is published here: the honest version of each engagement is the approach, not a number without a baseline.

Remote slope monitoring

An active slope required continuous evidence of movement across a wide footprint, through weather, with no power available at the instruments. The architecture pairs vibrating-wire piezometers with inclinometers on low-power loggers, uplinked over long-range radio to a gateway with cellular backhaul.

Thresholds are set per instrument by the geotechnical engineer responsible for the asset. Raw channels are retained alongside processed values so an alert can be reconstructed later, and the dashboard leads with the two numbers the site team actually watches.

Instruments: Piezometers, inclinometers, load cells
Uplink: LoRaWAN to a cellular gateway
Power: Battery with solar assist at the gateway
Output: Threshold dashboard, daily summary, raw archive

UAV-assisted infrastructure inspection

An inspection programme was producing hours of footage per visit and days of manual review. We specified the platform and payload, then built an on-board pipeline that detects and frames candidate defects during the flight, so review starts from a tagged set rather than a timeline.

Repeat visits use the same flight geometry, which turns two inspections into a comparison instead of two separate films — the difference between looking at pictures and seeing a change.

Platform: Multirotor with stabilised imaging payload
Compute: On-board inference and defect framing
Workflow: Tagged review set instead of raw footage
Repeatability: Fixed flight geometry for change detection

Solar asset monitoring

Radiometric thermal capture across solar rows, with automated anomaly detection that converts a survey into a list of locations and severities a maintenance crew can act on. Faults stop arriving months late inside an image folder.

The output is deliberately plain: a table, a map and a severity ranking, exported into the workflow the operator already uses.

Sensor: Radiometric thermal imaging on UAS
Detection: Automated hotspot and anomaly flagging
Output: Work-order list with location and severity

Utility monitoring & connected data

Distribution assets monitored at the edge, with field evidence joined to location and pushed through APIs into the utility's own tooling. Crews are dispatched to a fault rather than a district.

The integration mattered more than the dashboard: the utility already had a system of record, and adding another portal would have guaranteed the data went unread.

Edge: Sensing and acquisition on distribution assets
Integration: APIs into existing utility tooling
Outcome: Located faults, fewer speculative site visits

A note on evidence

Client identities, site details and measured results are published only with permission. Where a figure appears elsewhere on this site it has been verified for that context; where it has not been verified, it does not appear.

Imagery on this site is illustrative. No picture on these pages should be read as a photograph of a client installation.

Discuss a comparable project