SeLoger / MeilleursAgents2020–2022
Replacing a frozen autocomplete with a living reference
8
AVIV Group sites served · ~10 ms responses on critical paths
France’s leading real-estate group (think Zillow). The legacy autocomplete was fast, but its giant in-memory cache made data updates nearly impossible.
Raw OpenStreetMap data ingested continuously, then event-driven read models (Elasticsearch, PostgreSQL, in-memory .NET) instead of a frozen cache.
Read the case study →