PRODUCTION SCENARIO
A biotech company wants to fine-tune a publicly available open-weight language model on its own lab notes and deploy the result to an endpoint in its own AWS account. The company does not want to assemble training infrastructure from scratch.
Which AWS capability will meet these requirements?
Answering here is anonymous. Nothing is saved unless you sign in.
Show answer and explanation
Answer: Amazon SageMaker JumpStart
JumpStart provides pretrained open-source models that you can incrementally train and fine-tune and then deploy to an endpoint, with the training infrastructure built into SageMaker Studio. Model Registry catalogs and versions models that already exist; it trains nothing. Custom Model Import brings in weights customized elsewhere, so it assumes the fine-tuning already happened. Knowledge Bases grounds answers by retrieving the lab notes at query time and never changes the model's weights.