Vedas Kudalkar·3d agoHow I Cut My LLM Calls in Incident Triage With a Sliding Window and a CacheThe Cheapest LLM Call Is the One You Don’t Make
Joey Yi Zhao·3d agoThe Definitive Guide to pgvector: Transforming PostgreSQL into an Enterprise-Grade AI DatabaseThe explosion of Large Language Models (LLMs) and autonomous AI agents has forced software engineers to confront a fundamental…
Erwin Hermanto·Oct 29Vector Search at Scale: Sharding pgvector for 100M+ EmbeddingsA few months back, our semantic search feature was humming along nicely on pgvector. Then we crossed roughly 15 million embeddings and…A response icon1A response icon1
Manoj Kumar·Oct 28Build a RAG assistant, Day 1: write the golden set, then start Postgres with pgvectorTL;DR
HobbiesPark·Oct 26Spring AI 2.0 RAG Missed 94 of 100 Filtered Searches — pgvector Had the DocsA docs assistant held 200 chunks for one library in pgvector, a user asked a question those docs answer, and Spring AI 2.0.1 sent the…
DhanushKumar·Oct 23PostgreSQL in Modern Software Engineering: From Database Fundamentals to Production-Grade…An end-to-end engineering guide to PostgreSQL architecture, SQL, ACID transactions, MVCC, indexing, performance, security, scalability…
_Prav·Oct 22What Production RAG Needs That the Demo Doesn’tLessons from running retrieval inside live phone calls, where the documents are untrusted, the index is shared and the caller is waiting
Erwin Hermanto·Oct 18Building a Semantic Layer on Top of PostgreSQL Using pgvector and LLMsA few months back, one of our product managers asked a question that sounded simple on paper: “Can we let non-technical people query our…
InTech and Me, Loving It or Hating ItbyThe AIExplorer·Oct 16Your AI Features Don’t Need a Dedicated Vector Database. They Need PostgresHow a failed deletion created a data leak, and what we’d have built instead
Muhammad Mozzammel Haque·Oct 15pg_vectorsearch benchmarked: 14x faster index builds, a tie with IVFFlat on queriesI compiled the alpha extension on Postgres 18.4, loaded 500,000 vectors, and compared it with pgvector’s IVFFlat and HNSW on recall…
