[Performance]: Parallelize Batch Embedding Generation

[Kavirubc]

Problem Statement

The EmbedBatch function used by the CLI index command currently processes embeddings one by one in a serial loop. This significantly slows down bulk indexing for large repositories.

Proposed Solution

• Refactor EmbedBatch in internal/integrations/gemini/embedder.go to use goroutines (and errgroup) to parallelize embedding requests.
• Since Gemini SDK supports concurrent calls, this will allow faster processing of chunks during bulk indexing.

Context

Identified during bulk indexing of the simili-bot repository.

[gh-simili-bot]

Simili Triage Report

Note

Quality Score: 8.5/10 (Good)
The issue is well-described.

Classification

Category	Value
Labels

Similar Threads

Similarity	Thread	Status
82%	#18 [v0.1.0] CLI: Implement 'index' command for bul...	Open
81%	#19 feat: Implement 'index' command for bulk issue ...	Open

---

_{Generated by Simili Bot}