Back

quirk

Go

RAG (Retrieval-Augmented Generation) based data processing pipline that converts raw data into vector embeddings & Knowledge triplets for usage in vector databases

1 stars 1 forks 0 open issues

View on GitHub

Quirk is a production-ready fullstack AI pipeline: automated corpus ingestion, embedding generation, vector search, and scalable API delivery. Built for reliability and observability with Go services, Postgres-backed auth, and Dockerized deployment pipelines.

  • RESTful API for data submission, similarity search, and automated retrieval.
  • Multi-format support for .pdf, .csv, .json, .md, .xml, and .yml.
  • Cloudflare Workers AI integration with @cf/baai/bge-large-en-v1.5 (1024 dimensions).
  • Export capabilities to JSON, CSV, and direct ChromaDB vector integration.
  • JWT-based authentication with persistent PostgreSQL token storage.
  • Automated deployment architecture through Dockerized microservices and CI/CD.