Nearest-neighbour search over a billion embeddings: HNSW, IVF-PQ and disk graphs, filters, segments, sharding, and hybrid search with re-ranking.
Data engineers building storage, streaming and processing systems.
Your approach: Use diagrams or prose for data contracts, state ownership, throughput, ordering and recovery where the brief requires them.
Design a Vector Database. Nearest-neighbour search over a billion embeddings: HNSW, IVF-PQ and disk graphs, filters, segments, sharding, and hybrid search with re-ranking. Work from the scoping questions below. State assumptions for any unspecified load, guarantee or target, then trace your design end to end. Explain one difficult case and a credible alternative; the worked example is a reference, not a required implementation.
Resolve the scoping questions for a Vector Database. Separate stated behavior from assumptions, and identify what is outside your design.
Declare relevant volume, latency, freshness, quality or cost targets with units. Show calculations or an evaluation plan that can test them; unspecified targets are your assumptions, not hidden pass criteria.
Explain how your guarantees hold in a difficult case relevant to this subject. Address: Filters? Does cost matter?
Identify users, required behavior and exclusions. Answer: How many vectors and collections? Traffic?
Define the information owned by the system and the inputs, outputs and errors at its boundaries. Resolve: Recall and latency? Keyword search too?
Estimate the dominant workload and resource demand with units and explicit assumptions. For a learned system, also state how quality is measured and what data is available.
Draw or describe the responsibilities needed for a Vector Database. Trace a representative request, event or job from its input to a visible result; identify durable state owners.
Walk through a difficult case step by step, including detection and recovery. Consider: Freshness and consistency? Filters? Does cost matter?
Compare a credible alternative using your chosen workload and guarantees. Explain a remaining risk, a signal to watch and when you would change the design.