root 04770c97eb HNSW vector index: 100K search in 27ms (58x faster than brute-force) ...

- instant-distance HNSW implementation for approximate nearest neighbors
- HnswStore: build from stored embeddings, in-memory index, thread-safe
- POST /vectors/hnsw/build — build index from Parquet (100K in 35s release)
- POST /vectors/hnsw/search — fast ANN search
- GET /vectors/hnsw/list — list loaded indexes

Benchmark (100K × 768d, release build):
  Brute-force: 1,567ms
  HNSW:           31ms (50x)
  HNSW warm:      27ms (58x)

Build cost: 35s one-time for 100K vectors (release mode)
ef_construction=40, ef_search=50 — good recall/speed balance

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-27 20:00:50 -05:00

..

chunker.rs

Phase 7: Vector index + RAG pipeline

2026-03-27 08:12:28 -05:00

hnsw.rs

HNSW vector index: 100K search in 27ms (58x faster than brute-force)

2026-03-27 20:00:50 -05:00

index_registry.rs

Phase 11: Embedding versioning — model-proof vector layer

2026-03-27 09:27:10 -05:00

jobs.rs

Background job system for embedding — fixes 100K timeout

2026-03-27 09:03:07 -05:00

lib.rs

HNSW vector index: 100K search in 27ms (58x faster than brute-force)

2026-03-27 20:00:50 -05:00

rag.rs

Phase 7: Vector index + RAG pipeline

2026-03-27 08:12:28 -05:00

search.rs

Phase 7: Vector index + RAG pipeline

2026-03-27 08:12:28 -05:00

service.rs

HNSW vector index: 100K search in 27ms (58x faster than brute-force)

2026-03-27 20:00:50 -05:00

store.rs

Phase 7: Vector index + RAG pipeline

2026-03-27 08:12:28 -05:00

supervisor.rs

Dual-pipeline supervisor for embedding ingestion

2026-03-27 09:06:28 -05:00