Led architecture rewrite for user incentive program from scheduled job to real time stream processing - authoring design, building out features, and mentoring other engineers, on a short timeline.
Use Mongo transactions to handle atomic updates for high throughput concurrent processing.
Use Faktory-type service to coordinate jobs and handle retries.
Built SQL derived table pipeline to calculate incentive amounts with complex requirements such as
campaign attribution based on prioritization logic, multi-campaign cascading, and transaction ordering.
Enhanced price alert notifications after AI insights finished generating.
Built out notification for users when their recent trade value increases past a threshold.
