Datakimia

Turn a data warehouse into a product customers sign in to

Datakimia's clients had modelled data in a cloud warehouse, but getting it in front of their own customers still meant manual reporting. The portal had to change that: a site each client could brand as its own, where people sign in with their work account and see only the dashboards, and the rows within them, that they're entitled to. The hard part was trust. Users, roles and data permissions had to line up exactly between the portal and the analytics engine underneath it, for every tenant.

I started with a proof of concept that trialled two embedded BI options before we standardised on open source Apache Superset. I then set up the production front end in Next.js and wrote nearly all of it: SSO with Google and Microsoft work accounts, embedded dashboards with access granted by the server, user and role administration, and a themeable design system so each client's portal carried its own brand. I also contributed to the NestJS API that brokers access to the BI layer.

Leading the team meant setting how we worked as much as what we built: repository setup, pull request templates, CI with end to end tests on every preview deploy, and feature flags with expiry dates so unfinished work could ship hidden and never linger. I also did front end work on client projects, including most of a major Angular version upgrade and a Next.js analytics site I set up.

• 4Team repositories I set up from scratch
• 80+Front end pull requests merged into the portal
• 8Product and client codebases I shipped to