Search, retrieval and Elasticsearch
A reading path through Shivam Agarwal’s writing on vector search, hybrid retrieval, Lucene segment merges, and Elasticsearch storage tradeoffs.
This collection brings together my writing on search, from keyword and vector retrieval to Elasticsearch internals. The linked writing pages contain short introductions; the complete explanations and examples are on Medium.
Which article should I read first?
- Explore retrieval approaches: begin with vector search, then hybrid search, which covers combining keyword and vector retrieval.
- Explore Elasticsearch internals: read Lucene segment merges and StoredFields versus DocValues.
Where does retrieval connect with AI?
RAG: Giving Your AI a Library Card introduces retrieval-augmented generation. The AI agents collection continues into agent memory and coding workflows.
For how these subjects fit into my professional background, see technology strategy and engineering work.
Writing in this collection
Elasticsearch: StoredFields vs DocValues — The performance pitfall
The performance tradeoffs between stored fields and doc values in Elasticsearch.
July 12, 2025 · On MediumSearch: Vector Search — Beyond Keywords to Meaning
Moving beyond keyword matching with vector search.
July 9, 2025 · On MediumSearch: Hybrid Search — Unlocking Smarter Results
Combining keyword and vector retrieval with hybrid search.
July 9, 2025 · On MediumElasticSearch: Navigating Lucene Segment Merges
Understanding Lucene segment merges in Elasticsearch.
July 8, 2025 · On Medium