nethra0906/ClusterCache
A semantic search system using vector embeddings, fuzzy clustering, FAISS indexing, and a custom semantic cache with a FastAPI service.
What's novel
A semantic search system using vector embeddings, fuzzy clustering, FAISS indexing, and a custom semantic cache with a FastAPI service.
Code Analysis
13 files read · 4 roundsA FastAPI semantic search service over the 20 Newsgroups dataset that uses Sentence Transformers embeddings, FAISS exact cosine-similarity search, and a GMM-partitioned semantic cache to avoid redundant query computations.
Strengths
Clean modular architecture with clear separation of concerns (embeddings, vector store, clustering, cache, API), a thoughtful model singleton pattern, and a genuinely clever cache design that uses GMM cluster prediction to partition the search space before doing cosine similarity within a bucket.
Weaknesses
Zero test coverage, the 'fuzzy clustering' is overhyped (GMM probabilities are computed and saved but never surfaced in the API or used beyond bucket partitioning), and the project is essentially a thin orchestration layer over well-known libraries (FAISS, Sentence Transformers, sklearn GMM) with no novel algorithmic contribution.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
10
Contributors
1
Files
27
Active weeks
1
Repository
Language
Python
Stars
1
Forks
0
License
—