perf(cloud-ai): AWS Rekognition blocks the event loop on 14 synchronous boto3 calls

[groupthinking]

Problem

AWSRekognition (src/youtube_extension/integrations/cloud_ai/providers/aws_rekognition.py) issues 14 synchronous boto3 calls directly inside async def methods. boto3 has no async API, so every one of these blocks the event loop for a full network round-trip.

Method	Blocking boto3 calls
_test_connection	describe_collection
analyze_image	detect_labels, detect_faces, detect_text, detect_moderation_labels
get_service_status	describe_collection
_start_video_analysis	start_label_detection, start_face_detection, start_text_detection, start_content_moderation
_wait_for_job_completion	get_label_detection, get_face_detection, get_text_detection, get_content_moderation

_wait_for_job_completion is the worst case — it polls on a 5s interval up to max_wait_time (default 600s), so a single video analysis can stall the loop up to 120 times.

_prepare_image_input additionally does open(image_url, 'rb').read() on the loop for local files (the sibling http(s) branch already correctly uses httpx.AsyncClient).

Reachability evidence

youtube_extension.integrations.cloud_ai.providers.aws_rekognition is in the transitive import closure of the primary production entrypoint (youtube_extension.main:app, root Dockerfile:93):

• backend/cloud_ai_routes.py:89,141 maps the aws_rekognition provider
• integrations/cloud_ai/integrator.py:297-298 imports and constructs AWSRekognition
• main.py:171 mounts cloud_ai_routes

Acceptance criteria

• All 14 boto3 calls dispatch through await asyncio.to_thread(...)
• The local-file read in _prepare_image_input runs off the loop
• All 89 pre-existing tests pass without modification
• New tests prove the loop stays responsive, and fail against the pre-change code
• ruff at parity with main

[linear-code]

GRV-231

[coderabbitai]

🔗 Related PRs

#224 - Empirical PoC: 10 AI/workflow videos through the pipeline + fix dead AI extraction path [merged]
#438 - fix(cloud-ai): unblock event loop in async OCR + PR remediation scan [merged]
#496 - fix(youtube): complete youtube-transcript-api >=1.0 migration across remaining call sites [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!