Claims platform for deposit-replacement insurance in five weeks
A spec in August became a claims platform in production five weeks later, built on a live platform I had been releasing since March.
- Sector
- UK rental deposit-replacement insurance
- Role
- Release owner for the live platform and lead engineer on the claims platform.
- Timeline
- March to September 2026
- Stack
- SvelteKit, TypeScript, AWS, Stripe
Outcomes
- 5 weeks From the written spec to the claims platform in production.
- 27 of 27 Production releases since March, all merged by me.
- 881 Unit and integration tests on the offline-first video evidence recorder.
- 0 Objects in the off-site backup store when I found backups had silently stopped. I restored them and verified a dump.
- 46% Of all commits on the platform over the same period.
The business problem
The client sells deposit-replacement insurance to UK renters. When a tenancy ends and a landlord makes a claim, that claim has to be filed, evidenced, reviewed and valued. The claims platform handles that from filing to valuation.
What I built
The spec arrived in the middle of August. Five weeks later the claims platform was in production, covering:
- a filing wizard for claims
- an adjuster review queue
- responses from tenants
- video evidence, recorded offline-first so a weak signal on a property visit doesn’t lose the footage
- valuation, from a rate card with AI assistance
The client was very positive at the demo, and they plan to train their sales staff to sell it.
Keeping the live platform healthy
Most of my time on this client was maintaining the platform the claims work sits on. I merged every production release since March, and I went looking for failures nobody had flagged.
The biggest was the off-site backups. They had stopped without anyone noticing, and the off-site store was empty. I got them running again with two scheduled functions and verified a dump, so the client knows the backups can actually be used.
I also put sanctions and anti-money-laundering screening for tenants and agents into production in March, fixed server errors on the admin commission and agent pages, and added a safety check to database migrations so the same class of failure couldn’t ship again.
What a client gets from this
A new product in production on a short timeline, on a platform where the operational basics are checked rather than assumed: backups that restore, releases with one owner, and tests where the product is hardest to get right.
Have a prototype that needs to become a product?
Tell me what it is and when it needs to ship. Email is the quickest way to reach me.