Hackathon Challenge — AI Quality Control for Trench Documentation | Sustainista
Hackathon Challenge · 2026

AI Quality Control for Trench Documentation.

Build a functional AI prototype that checks trench photo documentation, maps coverage gaps and flags construction risk before acceptance.

The Challenge

What we cannot see will cost us.

Fiber trench documentation is high-risk when survey data is wrong, photos are missing or burial depth is unclear.

Bad documentation can invalidate warranty claims, shift liability and create expensive fiber cuts during future excavation or road works.

Your task is to build an AI-powered quality-control prototype that reviews trench documentation and identifies where evidence is complete, weak or missing.

Every unreviewed trench section is a liability waiting to be triggered.
Resources

Start here.

The challenge is built around trench photos, route geometry and automated quality-control logic.

Image Data

Trench Photos

Use the provided route photo set as the main input for AI-based documentation review.

Route Data

GeoJSON Route

Map photo evidence to route segments and identify missing or weak coverage.

QC Logic

Compliance Signals

Check GPS/date metadata, duct visibility, sand bedding, pipe end seals, ruler readability and privacy issues.

Output

Map + Report

Produce a reviewer-friendly result: green, yellow and red segments plus a concise summary report.

01

Ingest

Load trench photos and route geometry into your pipeline.

02

Geo-match

Match image metadata to route segments, ideally at fine segment resolution.

03

AI review

Score photo quality and documentation evidence: ducts, bedding, depth, coverage and compliance.

04

Classify

Flag each segment as complete, partial or missing evidence.

05

Report

Show an interactive map, risk list or PDF-style report that a reviewer can act on.

Prototype Requirement

Fully functional AI-QC workflow.

Your prototype must process real or provided trench documentation inputs and generate a usable quality-control result.

It does not need to be production-ready, but the core workflow must function end-to-end: image data in, AI/QC logic applied, risk output generated.

Manual review should become faster, more complete and easier to defend.

No pure slide concept. Build the review workflow.
What to Detect

Turn photos into risk signals.

Your AI review can focus on a few high-value signals rather than solving every defect at once.

01
Coverage

Missing Evidence

Where route sections have no usable photo or survey evidence.

02
Quality

Poor Documentation

Where photos exist but ducts, bedding, ruler, seals or context are unclear.

03
Risk

Acceptance Hotspots

Where missing evidence should block sign-off or trigger follow-up.

Business Ideas

Ideas in the room.

These are starting points, not a menu. Build a focused solution around one clear reviewer, risk or decision.

Construction Acceptance

Automated sign-off assistant

Pre-check whether trench documentation is complete before a section is accepted.

Contractor Management

Evidence-based accountability

Flag missing or weak documentation so issues can be resolved while the trench is still actionable.

Network Risk

Future excavation risk map

Identify undocumented sections that could cause expensive fiber cuts later.

Operations

Reviewer productivity cockpit

Reduce manual review from days to minutes by prioritizing only problematic segments.

Warranty

Audit trail generator

Create structured evidence for warranty claims, compliance checks and contractor disputes.

Scale

Rollout QC platform

Design a repeatable workflow that can be used across future construction phases.

Tech Check

Saturday late afternoon / evening.

The Sustainista team will run a first technical check to see how far the AI workflow, route matching and output logic have progressed.

Show progress

Working pipeline

Be ready to show what data you ingest and how your first checks work.

Show output

Early risk result

Even a rough green / yellow / red segmentation is useful if the logic is clear.

Expected Deliverables

Show the workflow, not just the idea.

Each team presents a functional AI-QC prototype for trench documentation.

01
Prototype

AI-QC Workflow

Process trench photos and produce quality-control classifications.

02
Output

Map or Report

Show where evidence is complete, partial or missing.

03
Business

Value Case

Explain who uses it, what risk it reduces and why it should scale.

Find the risk before it becomes expensive.

Use trench photos. Match the route. Flag missing evidence. Protect the network.

Start with Resources →