← All work
Product · 2024

Daily Product Backend, CDC Feeds & Recommendation (Debezium + Gorse)

An influencer-marketing media-intelligence platform

Overview

The backend for the platform’s “Daily” product, a personalised daily-digest service combining change-data-capture (Debezium), a Gorse recommendation sorter and a Node.js/TypeORM data layer.

The Challenge

A daily product needs fresh, personalised content per user, which means streaming database changes in real time and ranking items by relevance, not just serving a static feed.

What We Built

A multi-service backend orchestrated with Docker Compose / Kubernetes (docker-compose.yml, kube.yml). It includes a Debezium-based CDC setup (debezium/) to stream PostgreSQL changes, a Python gorse-sorter/ service wrapping the Gorse recommendation engine to rank items, TypeORM data sources and migrations (ormconfig, migrate.sh, seeds/), and Node services for search, redirection and route indexing (search.js, redirector.js, routeIndex.js).

Technologies & Approach

Debezium captures DB changes for low-latency freshness; Gorse provides recommendation ranking; TypeORM manages schema and migrations; everything containerised for deployment on the platform’s Kubernetes infra.

Outcome / Impact

Delivered the data and recommendation backbone for the Daily product, pairing real-time CDC with personalised ranking to drive the web and mobile front-ends.

Capabilities Demonstrated

  • Building change-data-capture pipelines with Debezium
  • Integrating recommendation engines (Gorse) into product feeds
  • Containerised, migration-driven Node/Python backends
More work See all →