Database migrations for legal and insurance clients

The project

At Tessaract.io, a legal-tech startup, I looked after the production PostgreSQL databases behind its platform for law and insurance firms. I wrote PL/pgSQL and Python scripts to keep them fast, built reports, and migrated more than 300,000 records from over ten law firms onto the platform.

Working with clients

The work was client-facing too. On one onboarding, a large law firm’s accountants were unhappy about leaving the workflows they knew, and the launch date was close. Saying yes to every request would have sunk the launch, so I told them plainly what we could change before launch, what would come later, and why. I also wrote guides for the problems I expected them to hit, like reconciling records and handling missing fields. The onboarding went smoothly.

Looking back

On another migration, the client kept their real data back until close to delivery, for confidentiality reasons. We’d written the scripts against our guesses about its format. The real files had inconsistent column names, missing values and formatting we hadn’t planned for, and we had to clean and revalidate everything in the last few days.

Now I ask for a representative sample at the start, even an anonymised one, and write scripts that expect the data to be messier than promised.