Boto3 / MLflow model logging via temporary AWS credentials

Question

We are building an ML tracking service using MLflow as a backend. One issue we've run into is that in order to log models via MLflow's python API​, the user needs to have AWS credentials configured on their machine. Since our service is outward-facing, we can't really let users have the access key for our S3 bucket. Is there a mechanism for authenticating a boto3 client used by MLflow via some temporary AWS credentials? We can generate a signed URL with write permissions to the bucket, but it's unclear how we would then pass it onto the boto3 client / MLflow python api. Or can we do something with environment variables? In any case, if someone knows of a good way to do this - I'd greatly appreciate the help. Best, SP