• Projects
  • Contact
Back to projects

AI Word Similarity

Real-time word similarity API over 1.2M+ GloVe embeddings, served from a Redis vector store via Rust/Axum.

GitHub
rustaxumtokioredisglove

A real-time AI word similarity service built on GloVe embeddings (1.2M+ words), stored in a Redis vector database via asynchronous pipelined HSET streaming.

Highlights

  • Low-latency similarity queries exposed through a Rust Axum API, supporting multiple embedding sets/datasets.
  • Entire environment orchestrated with Docker Compose for reproducible development and deployment.
  • Batched 500 embeddings per pipeline, cutting database write times by over 200% on large datasets.