What I worked on
The work was interface and reconciliation focused: moving data between core and downstream systems under controls that assume, correctly, that everyone will eventually need to prove what happened. Every flow had to be able to account for itself — what it read, what it wrote, what it rejected and why — with a trail that would survive an audit rather than a code review.
That also meant a slower, more deliberate change process than I was used to. Segregated duties, controlled promotion, and evidence attached to every release. Frustrating at first, then genuinely instructive: most of the ceremony exists because someone once skipped it.
Systems & platforms
Representative work
Controlled batch interfaces
Scheduled interfaces between core and downstream systems, with checksums, control totals and rejection handling designed so a partial run could be proven partial rather than assumed complete.
Reconciliation and exception reporting
Reporting comparing what was sent against what was accepted, with exceptions itemised for the operations team rather than aggregated into a single pass or fail.
Audit-ready logging
Structured logging at the flow level capturing enough detail to reconstruct any given run after the fact, without carrying sensitive payload data into the log store.
What the sector taught me
Six months was enough to permanently raise my baseline. Banking treats “can you prove it?” as a design requirement rather than an afterthought, and I have carried that into every sector I have worked in since. It is the sector I would most like to go deeper in.