InTowards AIbyJames Luan·17h agoHybrid Search Starts with Two Good Candidate ListsConsider a support query containing both an error code and an ordinary description of what went wrong. Exact matching helps find the code…A response icon1A response icon1
InDesi StackbyYogesh Haribhau Kulkarni (PhD)·Oct 25Ask Your Own Notes, OfflineA Plain-Python Retriever and a Small Model
Lekha Priya·Oct 24AI Engineering Series Part 4: Hybrid Search with BM25 and Vector RetrievalWhat happens when semantic similarity alone can’t find an exact error code, ticket number, or config key, and why enterprise search needs…A response icon1A response icon1
VLM_Newbie·Oct 20[RAG] DPR: Dense Passage Retrieval for Open-Domain QA논문: Karpukhin et al. 2020, arXiv:2004.04906 (EMNLP 2020)
Samuel·Oct 18Lakebase Search BM25 vs Postgres ts_rank: the keyword step of a payer search, measured on throwaway…Lakebase BM25 against built-in Postgres ts_rank on 48,283 search keys from a 3,015,080-row payer table, measured on 8 throwaway branches of…
Shivansh Kotnala·Oct 11From RAG to Tool-Using AI SystemsRetrieval, reranking, grounding, structured tool calls and security boundaries
Sidhanttiwari·Oct 3Building a RAG Pipeline That Actually Works on Real DocumentsIn Part 1, we covered the architecture and stack decisions. This part covers the retrieval pipeline — specifically how to make it work on…
Indraneel Chatterjee·Oct 3Beyond the Hype of Pure Embeddings: Reconciling Lexical Precision (BM25) and Semantic Space for…Why production search engines and RAG systems in 2026 run hybrid retrieval — and the measurable business value it generates
Mehmet Çakmak·Oct 1The Evolution of Modern Search Systems: From SQL LIKE to Agentic SearchTürkçe: https://www.mehmetcakmak.online/blog/modern-searching-systems-sqllike-agenticsearch?lang=tr
Aditya J·Sep 21Scaling RAG Infrastructure: The Journey from Flat Search to DiskANNThis is the Part 1 of the scaling series
![[RAG] DPR: Dense Passage Retrieval for Open-Domain QA](https://miro.medium.com/v2/resize:fill:160:106/1*WdxDsYnvGqygP0QB2JrPYw.png)
![[RAG] DPR: Dense Passage Retrieval for Open-Domain QA](https://miro.medium.com/v2/resize:fill:320:214/1*WdxDsYnvGqygP0QB2JrPYw.png)
