Your engineering blog post on moving settlement off the synchronous fan-out is the reason I am applying. I spent last year on the same problem at Northwind, on a ledger that could not go past 400 writes per second without reconciliation drift, and the outbox pattern you described is very close to what we landed on.
I rebuilt that ledger on an append-only event store and took sustained throughput to 9,200 writes per second with exact balance reconciliation, then led the shard migration that moved 42 million accounts off a single Postgres primary in six weeks with no downtime and no rollback. The part I am proudest of is not the number: it is that we shipped it incrementally, behind a dual-write and a comparison job, so that a rollback was always one flag away.
I am looking for a team where payments correctness is the hard part of the job rather than an afterthought, and where I can keep owning a system end to end: design, rollout and the pager. I would welcome the chance to talk about how you are handling reconciliation at your current volume.
Thank you for your time, Priya Raghunathan