Introduce a specialized block only after the requirement creates its problem#
| Requirement | Consider |
|---|---|
| Keyword text retrieval | Inverted index / search engine |
| Ranked prefix suggestions | Trie/FST-style or precomputed autocomplete index |
| Nearby entities | Geohash or spatial index plus exact filtering |
| Scheduled execution | Scheduler, ready queue, idempotent workers |
| Top scores / hot counter | Ordered score structure / sharded aggregation |
| Web crawling / media | Frontier pipeline / derived-artifact jobs |
| Approximate membership / distinct / frequency | Bloom / HyperLogLog / Count-Min Sketch |
| Concurrent document editing | OT or CRDT family after defining semantics |
| Semantic similarity | Embedding pipeline and vector index |
| Low-latency predictions | Model serving; feature store only when justified |
Continue into vendor-specific detail only when the prompt or role requires it. The unpublished Technology Deep Dives remain a separate phase.
Related practice problems
Use the manifest-backed prompts below to apply this lesson in a complete design.
- Clusters, nodes, and shardsElastic Documentation
- Probabilistic data typesRedis Documentation
- Vector Search overviewGoogle Cloud Documentation
Reach this point to mark the lesson as read.