Skip to content

Remove deprecated functionality: FAISS embeddings, file backend, and EFS references #452

@aarora79

Description

@aarora79

Summary

Remove deprecated functionality that is no longer needed to reduce codebase complexity and dependencies.

Items to Remove

1. FAISS-based Embeddings Search

  • Remove FAISS vector search implementation
  • Remove faiss-cpu dependency from pyproject.toml
  • This will significantly reduce the virtual environment size (~38 MB for faiss + related libs)

2. File-based Backend

  • Remove file-based storage repository implementation
  • The project has moved to MongoDB/DocumentDB as the primary storage backend

3. EFS References from Terraform

  • Remove all EFS (Elastic File System) references from Terraform infrastructure code
  • EFS was used for the file-based backend which is being deprecated

Benefits

  • Reduced dependencies: Removing faiss-cpu, torch, sentence-transformers could save ~6GB in the virtual environment
  • Simpler codebase: Less code paths to maintain
  • Clearer architecture: Single storage backend (MongoDB/DocumentDB) instead of multiple options

Files to Review

  • registry/repositories/file/ - File-based repository implementation
  • registry/embeddings/ - FAISS and sentence-transformers usage
  • pyproject.toml - Dependencies to remove
  • terraform/ - EFS module and references

Acceptance Criteria

  • All FAISS-related code removed
  • File-based backend code removed
  • EFS Terraform resources removed
  • Dependencies updated in pyproject.toml
  • All tests pass after removal
  • Documentation updated to reflect changes

Metadata

Metadata

Assignees

Labels

architectureArchitecture and design decisionstechnical-debtTechnical debt that should be addressed

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions