3.3
3.3 Retrieval, Search & Vector Index Infrastructure
Inverted indexing, BM25 lexical search, HNSW/IVF-PQ vector engines, hybrid search (dense+sparse), index freshness, and ACL-aware retrieval infrastructure.
Inverted Index, BM25 & Enterprise Search Engines
Distributed inverted indices in Elasticsearch/OpenSearch, BM25 scoring, tokenization, sharding, query understanding, ACL security filtering, and offline relevance evaluation (nDCG/MRR).
🏢 Companies
Elasticsearch、OpenSearch、Vespa、ClickHouse
🛠️ Tech Stack
Inverted IndexBM25ElasticsearchOpenSearchVespaShardingACL RetrievalnDCG
💼 Roles & Salary
Search Engineer、Relevance Engineer、Retrieval Engineer
💰 $185K - $380K / year (Search & Relevance) | ¥450K - ¥1.1M / year
📚 Prerequisites: Information Retrieval & Inverted Index • BM25 Ranking Algorithm Derivations • Distributed Sharding & Query Caches • Offline IR Metrics (nDCG, MRR, Recall)
Vector DB & High-Dimensional HNSW / IVF Indexing
HNSW graph indexing & heuristic pruning, IVF-PQ product quantization, SIMD/GPU distance acceleration, hybrid scalar-vector filtering, and millisecond index freshness.
🏢 Companies
Pinecone、Zilliz / Milvus、Qdrant、Weaviate
🛠️ Tech Stack
Vector DatabaseHNSW GraphIVF-PQEmbeddingPineconeMilvus / ZillizQdrantHybrid Retrieval
💼 Roles & Salary
Vector Database Engineer、Retrieval Engineer、Search Engineer
💰 $190K - $395K / year (Vector Database Systems) | ¥480K - ¥1.2M / year
📚 Prerequisites: ANN Search Algorithms (HNSW, IVF-PQ) • High-performance C++/Rust & SIMD Ops • Product Quantization (PQ/SQ) Compression • Filtered Vector Search Execution Engines