Third Estate Ventures · Real-estate investment · Data & UX · 2025 – Present
Rebuilding the property data analysts decide with
Expanded usable property coverage from about 35K to 49K parcels and cut duplicate records by 60% by rebuilding how five municipal data sources were collected and validated, starting from how analysts actually make decisions.
- Role
- Business Intelligence & UX Designer, from analyst discovery through implementation.
- Timeline
- 2025 – Present
- Tools
- Python, REST APIs, Snowflake, SQL, Power BI, Power Automate
Context
Third Estate Ventures' analysts evaluate real-estate investments. Their decisions are only as good as the property data underneath, and that data comes from public sources that were never designed to work together.
The problem
Coverage stopped at roughly 35K parcels, and about 12K duplicate records meant analysts could not trust that one row was one property.
My responsibility
As a Multi-Project Fellow I worked across the whole path: analyst discovery, UX, data design, and implementation.
Approach
- Started with analysts: how they actually evaluate a property and which data they rely on.
- Rebuilt collection and validation for five municipal data sources in Python using their REST APIs.
- Modeled the cleaned data with SQL and Snowflake.
- Built Power BI analytics around analysts' real decision steps rather than around the raw tables.
What changed
Analysts now work from a validated, de-duplicated property dataset with far broader coverage, surfaced in views designed around their workflow.
Outcome
Coverage grew from about 35K to 49K parcels, and duplicate records fell by 60%, from about 12K to 4.8K.