Just a heads up: this project and its details are protected under an NDA. Please email me at hello@connormorgan.io for access.

Overview

A nonprofit research organization needed a way to locate and contact thousands of families displaced by a mid-20th century urban redevelopment project. Decades of historical records existed only as scanned PDFs — unstructured, handwritten, and spread across thousands of files. The team had no reliable way to search them, track investigator progress, or manage outreach at scale.

Context

The organization had been working from a manually maintained spreadsheet that couldn't scale. Records were incomplete, investigator assignments were tracked informally, and there was no centralized view of project status. As the archive grew and the team expanded, the existing setup created bottlenecks and made quality control difficult.

The client had also begun building an Airtable interface to manage outreach, but it needed to be rebuilt around a new, cleaner dataset and extended with features the team had identified through months of active use.

My Role

I owned the project end to end across two phases: data engineering and interface design. There was no developer. I built the pipeline, cleaned the data, designed the base schema, and built out the full Airtable interface.

Tech Stack

Python, Google Cloud Storage, Anthropic Claude API, Airtable, Microsoft Excel