Arize AI
Search…
Setting Up a File Import Job
How to set up an import job to ingest data into Arize
A few steps need to be completed for Arize to start syncing your files as model inferences
  • Create a bucket (if you don't have one already) and folder (optional) from which you would like Arize to pull a particular model's inferences.
    For example you might setup an S3 bucket and folder named s3://arize-models/cc-fraud/ that contains CSV files of your model inferences.
  • Navigate to File Import section of the your workspace in the Arize platform.
Navigate to the File Imports page
  • Create a file import job. And follow the steps to create a file import job.
  • Grant Arize access privileges. Copy the policy supplied by Arize in the file importer job setup and apply this policy to the bucket setup in the first step. E.g. s3://arize-models. For details on the policy, please see the bucket policy section.
Capture the policy to apply to the bucket
Add policy to your bucket
  • Add Proof of Ownership to your Bucket. Tag your bucket with the key set to arize-ingestion-key and the value set to the provided tag value. Ex: AWS Object Tags
Capture your unique arize-ingestion-key
Add arize-ingestion-key as a bucket tag
  • Define your file's schema in Arize to point to the bucket and prefix where files are or will be placed. Make sure to your file's column names are in compliance with Arize's model structure. See file schema
Define the schema of your files
  • Finish creating a file import job
Click finish to fire off the job
  • Add model data to the bucket under the prefix and ensure the columns match the model schema. Arize file import jobs manage workers that continuously keep track of your buckets so the new file will be discovered and processed during the next import interval.
Copy link