Series

Building Insight Engine

A build-in-public series documenting the end-to-end development of Insight Engine — a GraphRAG-powered cross-domain research discovery platform built on Neo4j, Qdrant, Mistral 7B, and FastAPI. From knowledge graph construction to cloud deployment, I'm sharing the architecture decisions, failures, and lessons as the project evolves.

Mar 23, 2026
Mar 23, 20265 min read25 views
Building Insight Engine: How I'm Using GraphRAG to Discover Cross-Domain Innovation (Part 1)
A deep-dive into building a B2B SaaS knowledge discovery platform — powered by 229K research papers, a knowledge graph, and a single RTX 4060 GPU.
5 min read25 views
Mar 25, 2026
Mar 25, 20267 min read4 views
The NLP Pipeline Behind Insight Engine: Extracting 10.7M Entities from 229K Research Papers (Part 2)
How I used spaCy, SciSpacy, and a rule-based relation extractor to extract 10,779,699 entities and 2,294,895 relationships from 229,498 arXiv papers — fully tracked with DVC and MLflow on a single RTX 4060 GPU.
7 min read4 views

Posts in this series

Building Insight Engine: How I'm Using GraphRAG to Discover Cross-Domain Innovation (Part 1)

The NLP Pipeline Behind Insight Engine: Extracting 10.7M Entities from 229K Research Papers (Part 2)