Plaid's APIs sit between thousands of apps and the banks behind them, so a slow or retried request can decide whether someone's payment goes through. That is the kind of backend problem I have spent eight years on, most recently at Stripe, and it is why your Senior Software Engineer opening on the Payments Platform team caught my attention.
At Stripe I cut p99 API latency 42% by redesigning the request-routing layer and adding adaptive load shedding, so traffic spikes degrade gracefully instead of cascading into timeouts. I then led the move of our billing services from synchronous calls to async workers. The hard part was not the queue itself but making every job idempotent, so a retry could never charge an account twice. After the cutover, throughput doubled to 18k requests per second. I also own multi-region failover for the payments path, which held 99.99% uptime through two regional outages.
Before that, at DoorDash, I built the dispatch service behind more than a million daily deliveries and trimmed assignment time 28%. I introduced contract tests across 12 services, which cut integration regressions 35% and made it safer for other teams to ship against our APIs. I work mostly in Go and Python on Kubernetes and AWS, and I mentor four engineers and run our backend on-call rotation and incident reviews. That work has taught me to write postmortems that change the code instead of just describing the outage.
Plaid depends on institutions it does not control, and designing for their failures is the part of this role I would most like to own. I'd welcome a conversation about your payments roadmap and am glad to walk through the billing migration design in detail.