Reasoning-Based Refinement of Unsupervised Text Clusters with LLMs

Create an LLM-as-a-Judge wrapper that cleans up existing unsupervised text clustering outputs. This allows developers to use standard clustering (K-means) and immediately refine it into meaningful, human-interpretable categories.

Suggested repo: cluster-clean

"Turn your noisy embeddings into clean, labeled topics."

Estimated effort: 30h