Understand this story
Exowatt built a reusable AI pipeline using Amazon SageMaker AI serverless model customization. The pipeline fine-tunes Qwen3-1.7B for edge alert triage and Qwen3-14B for cloud-based energy market analysis, combining LoRA and SFT, and scores each candidate model on 2,000 answers using SageMaker serverless evaluation. The fine-tuned Qwen3-14B is hosted through Amazon Bedrock Custom Model Import. Exowatt says this repeatable path from raw operational data to evaluated deployment artifacts can now be completed in days rather than months.
AI-generated from public reports
Latest development · Report
Exowatt Beats Frontier Models in Accuracy with SageMaker Serverless Model CustomizationDevelopments
1 confirmed developments
Exowatt builds edge and cloud model pipelines with SageMaker serverless fine-tuning
Source:AWS Machine Learning
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 28
AWS Machine LearningFirst-party
Exowatt Beats Frontier Models in Accuracy with SageMaker Serverless Model Customization
Using Amazon SageMaker AI serverless model customization, Exowatt fine-tuned Qwen3-1.7B for edge alert triage and Qwen3-14B for cloud-based energy market analysis, surpassing frontier models in accuracy while cutting costs by up to 49 times. The pipeline combines LoRA and SFT, and uses SageMaker serverless evaluation to score 2,000 answers for each candidate model; the fine-tuned Qwen3-14B is hosted through Amazon Bedrock Custom Model Import. Exowatt says this repeatable path from raw operational data to evaluated deployment artifacts can now be completed in days rather than months.