Right now I'm building Reforge, an in-memory key-value store built from scratch in Go, inspired by Redis and exploring how vector search and embeddings can be integrated into an in-memory datastore.
— Building
Reforge is an in-memory key-value store written in Go, implementing the RESP protocol from the ground up along with core Redis data types, persistence, and replication. It's extended with native vector similarity search, making it usable as a lightweight backend for AI/RAG applications.
github.com/RudraPratapDev/reforge↗— Exploring
Skein — local-first semantic memory for your AI conversations. The idea is to recall the right context across ChatGPT, Claude, and Gemini, powered by a vector database you own. Reforge might eventually serve as the storage layer underneath it.
github.com/RudraPratapDev/Skein↗