Engineering case study

Relay Console

A marketing and engagement console for a consumer-services app: WhatsApp campaigns, a live AI-assisted inbox, an in-browser creative studio, and hourly data pipelines feeding an analytics layer. One containerized Node service on Cloud Run. Pick a flow to walk through.

Anonymized: client name, project ids, secrets and domains removed. Diagrams describe architecture only and show no production data.

00System architecture

How the pieces fit: Cloud Run, BigQuery, Mongo, Cloud Tasks, GCS, Cloud Functions.

01WhatsApp campaign send

Audience to delivery: media cache, parallel fan-out, scheduling. ~10x faster.

02Data-sync pipelines

Meta + GA4 + Firestore to BigQuery, idempotent upserts, anomaly alerts.

03Chat + AI assist

Inbound message to reply: rate limit, retrieval, provider fallback, human handoff.

04Creative Studio

Generate, compose and publish: AI gateway, Chromium render, signed URLs.

05CI/CD to Cloud Run

Push to main: test, build, deploy, pin traffic to latest revision.

Relay - anonymized engineering case study. Code in ../src.