Data Engineering
CAD-to-Structured Data Pipeline.
Turning masterplan drawings into structured operational data
01
Overview
In many property and development workflows, a masterplan drawing is more than a visual document.
It contains the foundation of the project: lots, boundaries, dimensions, areas, attributes, and spatial relationships that downstream teams depend on.
But if that information remains trapped inside CAD drawings, teams end up recreating the same data manually across spreadsheets, maps, pricing models, development workflows, and sales systems.
This project focused on turning masterplan drawings into structured, reusable, operational data.
02
The operational challenge
The organization worked with masterplan drawings supplied in inconsistent formats and styles.
Historically, parts of this process involved printing masterplans, interpreting drawings manually, and maintaining project mix data in spreadsheets.
That workflow created a fragile foundation for downstream operations.
When lot-level data is manually created or re-entered across systems, every downstream function is affected — development, pricing, cost tracking, sales, discounts, enquiries, and reporting.
Lot boundaries.
Boundary polygons.
Lot dimensions.
Lot areas.
Lot attributes.
Spatial layout.
Classification details such as regular or corner lots.
03
Before
The old workflow relied heavily on manual interpretation and spreadsheet-based tracking.
Lot data had to be extracted manually from drawings.
Different suppliers produced CAD files in different structures.
Teams relied on spreadsheets as operational references.
Lot attributes and geometry could become inconsistent across systems.
Downstream teams did not always have a single source of truth from the beginning.
Manual handling increased the risk of errors and rework.
04
The AI opportunity
The opportunity was to use AI and geospatial processing to convert complex masterplan drawings into structured data that could support the full project lifecycle.
The goal was not only extraction.
The goal was to create a reliable lot-level data foundation.
Accept masterplan drawings from different suppliers.
Identify individual lots.
Extract each lot boundary as a polygon.
Calculate or validate dimensions and area.
Capture lot-level attributes.
Visualize the extracted data on a map.
Produce structured tables for downstream use.
Support validation where geometry or calculations were uncertain.
05
The solution
I helped build a CAD-to-structured-data pipeline that allows users to upload masterplan drawings and receive structured lot-level outputs.
The system extracts lot boundaries and attributes, converts them into tabular and geospatial formats, and visualizes them on a map.
The resulting dataset can then act as a reliable starting point for downstream workflows.
Instead of treating the drawing as a static artifact, the pipeline turns it into an operational data layer.
06
How the system works
Upload masterplan drawing
Users provide the source drawing.
Parse and interpret geometry
The system identifies lot structures, boundaries, and spatial relationships.
Extract lot-level data
Each lot is converted into a structured record with geometry, dimensions, area, and attributes.
Validate and normalize
The system handles inconsistencies across supplier drawings and applies checks to detect geometry or calculation issues.
Visualize
The extracted project mix is displayed on a map for review.
Publish downstream
Structured lot data becomes available for downstream development, pricing, sales, enquiry, and reporting workflows.
07
Trust and governance considerations
This project involved real-world CAD complexity.
The challenge was not simply reading a clean file. The system had to deal with inconsistent drawings from different suppliers, different conventions, and occasional geometry issues.
Variation in CAD drawing structure.
Supplier-specific drawing conventions.
Small geometry and calculation discrepancies.
Validation of lot dimensions and areas.
Review workflows for uncertain outputs.
Preventing downstream systems from relying on incorrect geometry.
Creating a reliable single source of truth from the start of the project lifecycle.
This is where human-in-the-loop validation matters. For high-value spatial data, AI should accelerate extraction, but validation is essential before downstream reliance.
08
Impact
The system changed the role of masterplan drawings in the operational workflow.
Instead of being manually interpreted and re-entered into spreadsheets, drawings could become structured data assets.
Reduced manual effort in extracting lot-level information.
Faster creation of complete project mix data.
Better consistency across development, pricing, and sales workflows.
Earlier creation of a lot-level single source of truth.
Reduced rework caused by manual spreadsheet handling.
Improved ability to visualize and validate project data spatially.
Stronger downstream analytics and reporting foundations.
The value was not just automation. The value was creating a data foundation that supports the full lifecycle of each lot from planning through to sales and performance tracking.
09
What this project demonstrates
This project demonstrates one of the strongest use cases for applied AI in enterprise settings: converting unstructured or semi-structured operational artifacts into reusable systems of record.
The business did not need another dashboard.
It needed a way to turn planning artifacts into reliable operational infrastructure.
Next Case Study