Fintech feature team mobilization

Context

A fintech scaling a lending/payments AI feature. Production data and compliance environment access are the ramp bottleneck, and, the sharper risk, the founding ML engineer is the single point of failure across most of the stack and is rolling off fast.

The decision

Fix the bus factor before the access: pre provisioning helps the ramp, but the existential risk is a founding engineer who's the only owner of the pipeline, feature store, and fraud rules, schedule KT with named backups now.

What most miss

Scale ups celebrate the hero engineer and miss that heroism is concentration risk. When one person owns the training pipeline, the feature store, and the fraud rules, their departure is an outage waiting to happen.

Stakes

Lose the founding engineer before KT and the fraud rules and training pipeline have no owner overnight.