AWS Bedrock Model Integration | Kamiwaza Docs

This is documentation for Kamiwaza 0.8.0, which is no longer actively maintained. For the current GA release, see 1.0.1.

Version: 0.8.0

Overview

Kamiwaza 0.7 adds first-class support for hosting models backed by Amazon Bedrock. Administrators can register a Bedrock model, deploy it through the existing model lifecycle, and expose the familiar /v1/chat/completions and /v1/completions APIs to application teams.

Platform assumptions

Register a Bedrock model

  1. Navigate to Models → Add Model.
  2. Select AWS Bedrock as the engine.
  3. Provide the model details:
    • Display name – friendly name shown in the UI.
    • Model identifier – Bedrock model ID (for example anthropic.claude-v2).
    • AWS Access Key ID and Secret Access Key – scoped credentials for invoking Bedrock.
    • AWS Region – currently validated with us-west-2.
  4. Save the model. It will appear in the catalog alongside local engines.

Deploy and test

  1. Open the newly created model and click Deploy. Kamiwaza provisions the routing entry; no containers are launched because Bedrock handles execution.
  2. Use the Test Conversation modal or your own HTTP client to exercise the /v1/chat/completions endpoint. Responses stream through Kamiwaza just like local engines, and the raw Bedrock payload is preserved for troubleshooting.
  3. Monitor the deployment from the usual dashboards—metrics and decision logs continue to flow through the platform observability stack.

Operational notes

Next steps