Model Build API, Athena Data Source Support & more - Product Updates 01/2024

We’re delighted to announce our latest product updates, including support for Athena Data Source, model build API, better usability for the Qwak Vector Store, and more. These updates are part of our ongoing efforts to improve the Qwak platform, amplify user experience, and streamline your work in one platform as your single source of MLOps truth.

Model Build API

Build models everywhere, directly from your notebook or from your local machine by simply running “qwak.build_model()”

1. Build models from Workspaces

Qwak Build SDK simplifies handling local or Jupyter notebook-based models. It ensures precise versioning, tracking, and effortless transition from research to production.

2. Python-driven model builds

Automate model builds using Python and seamlessly integrate with CI/CD pipelines and automation workflows.

3. Streamlined versioning for pre-built models

Build and register pre-trained models from any Python environment by supplying existing trained model instances, skipping remote build phases.

# Any Qwak model
model = MyQwakModel()

# Build your model remotely
QwakClient().build_model(
  model_id='llama2',
  prebuilt_qwak_model=model
)

Learn more →

Athena Data Source Support

Introducing AWS Athena batch source: Seamlessly configure connections to data-at-rest by creating a configuration object linking to raw data. With the new Athena source, Qwak connects to Amazon Athena, enabling users to query and ingest data, allowing users to consume features for training and low latency real-time serving based on the data from Athena. 

from qwak.feature_store.sources.data_sources import AthenaSource
from qwak.feature_store.sources.source_authentication import AwsAssumeRoleAuthentication
from qwak.feature_store.sources.time_partition_columns import DatePartitionColumns

athena_source = AthenaSource(
    name='my_athena_source',
    description='my Athena source description',
    date_created_column='date_created',
    aws_region='us-east-1',
    s3_output_location='s3://some-athena-queries-bucket/',
    workgroup='some-workgroup',
    query='SELECT * FROM "db"."table"',
    aws_authentication=AwsAssumeRoleAuthentication(role_arn='some_role_arn'),
    time_partition_columns=DatePartitionColumns(date_column_name='date_pt', date_format='%Y%m%d'),
)

Visit our documentation to learn how you can easily connect your Athena data source in minutes.

Vector Store Multi-Process Ingestion

With the latest enhancement in SDK version 0.5.41, users can now upsert vectors, allowing the addition of new ones or replacement of existing vectors with the addition of controlling the batch_size as the maximum number of vectors in a single backend request; multiproc, determining the use of multiprocessing during insertion; and max_processes, setting the maximum number of processes.

See docs →

Rebuild Models in One Click

Instantly rebuild your models in one-click directly from the model builds table or the individual build view. Reduce the time and steps required to rebuild models, streamline your workflow, and iterate quickly on your models. With the ability to initiate rebuilds directly where you are working, there's no need to switch contexts or move away to the terminal console or Jupyter notebook. Simply choose the required build, click Rebuild and you’re all set!

Model analytics got a new look

The model analytics tab just got a fresh new look! This upgrade brings features like easy sorting, filtering, and infinite scroll, with a revamped look for better usability. A key addition is the "Hide editor" button, tailored for smaller screens, allowing users to concentrate on the table view for a more focused analysis.

Do More With Qwak

If you’re interested in learning more about how Qwak can help you get your models to production faster and more efficiently, let's talk.