Skip to content
Kitbay

aws-ai-ml

Apache-2.0Updated on Nov 11, 20263,464GitHub starsIncludes scripts

An AWS skill covering model customisation and deployment on Amazon SageMaker, from planning to production.

What the skill teaches

An AWS skill covering model customisation and deployment on Amazon SageMaker, from planning to production. It routes the agent to reference folders for defining a use case, choosing a base model from SageMaker JumpStart or the Hub, preparing and checking datasets, fine-tuning with SFT, DPO, RLVR or RLAIF, evaluating results, deploying to SageMaker or Bedrock, optimising inference, diagnosing endpoints and managing SageMaker MLflow. Many folders include Python templates and helper scripts. Ground Truth labelling, Feature Store and general AWS infrastructure are outside its scope.

Capabilities

  • Plan a model customisation project on SageMaker
  • Select a base model from SageMaker JumpStart or the Hub
  • Prepare training data and run fine-tuning jobs
  • Evaluate models and deploy them to SageMaker or Bedrock
  • Diagnose SageMaker endpoint failures and latency