I’m stuck on what the approach should be for caching serialized objects as part of search queries. Right now, the serializers are performing extra queries and even with prefetch_related, a typical query is taking around 40 seconds to complete.