Back to projects

Data platform · Event-driven architecture

Self-Service Insights Platform

A GCP event-driven platform where subject-matter experts define heuristic processes on incoming data and trigger alerting workflows.

Pub/SubCloud RunSpanner

The challenge

Subject-matter experts often know which signals matter, but translating that knowledge into production data workflows can create a queue of engineering work. The goal of this platform is to let those experts define heuristic processes without giving up the operational controls expected of a production system.

The approach

Incoming events move through a GCP-based processing flow. Configured heuristics evaluate the data and can trigger downstream alerting workflows when their conditions are met.

The platform separates three concerns:

  1. receiving and distributing incoming events with Pub/Sub;
  2. running scalable processing services on Cloud Run; and
  3. storing configuration and operational state in Spanner.

Engineering focus

The design emphasizes self-service configuration, traceable processing, and clean boundaries between domain rules and the infrastructure that executes them. This makes it possible to evolve expert-authored heuristics while keeping ingestion and alert delivery dependable.