Building a personalised home feed for Germany’s biggest classifieds website
summary
Prototype stores simple user profiles on the mobile client and builds queries against the live search index but strain and complexity force a rethink for a platform with ~25 million live ads and a 70–80 person team. The production pipeline uses Kafka Streams with embedded state (served from RocksDB) to materialize real-time profiles, updating within seconds, scaling by adding instances per topic partition, and driving a big lift in engagement while warning about bleeding‑edge versions and careful Kafka configuration.
AI-written summary · about
chapters (4)
- 0:00Start
- 3:41Mission: inspire users!
- 9:46How can I start building a feed?
- 17:48kafka A distributed streaming platform
related talks
20mArchitecture of relevancy search at mobile.de
26mPersonalizing User Content - The Power of a Vector & Search DB
27mA smooth ride: Online car buying and selling at mobile.de
33mBuilding Recommendation Systems for Global E-commerce Marketplaces
29mBuilding ML Platform Capabilities at Global Scale
23m101 A/B Tests for personalized News
20mDataEngBytes 2023 - PER-T1-06 - Kamal Hossain
36mDigital, data-driven financial advisory