Choosing the right embedding strategy for FileNest

[lakshyajain-0291]

🤔 Thought: Which embedding strategy should FileNest adopt for initial MVP?

We’re debating between:

• Pre-trained Sentence Transformers (e.g., all-MiniLM)
• Custom-trained lightweight models
• Static TF-IDF embeddings
  ...and some others

Points to consider:

1. Performance overhead vs embedding quality
2. Model size and inference latency
3. Ease of integration for beginners
4. Upgrade path to multimodal embeddings (image/text/video)

Do read our docs in FileNest Docs

Feel free to share benchmarks, past experiences, or preferences!
Looking forward to hearing pros/cons for each approach. 🔍