Product Overview

Qwak Inference Lake is a fully managed data lake that comes as an off-the-shelf product with the Qwak Platform. In Qwak Lake you can find all your inference, feedback, and baseline data for each model and its different versions (aka builds).

The Qwak Inference Lake data can be accessed directly via the Qwak SQL interface on the Analytics page, or via any other query or data processor engine.

The Main Pillars

Data Observability

Track your models production data and compare it to that model’s training and its previous versions.


Log all your inference data from its inception, at any scale.

Performance Management

Monitor applicative and statistical model performance metrics.

Getting Started

Setting up Qwak Inference Lake is simple, requiring only one line of code. All you need to do is import the Qwak SDK and wrap the relevant functions (the predict function, in our example) with @analytics, and that’s it—from now on, Qwak will automatically track and manage all your data into Qwak Lake.

Qwak Lake Use Cases

Track and monitor your models features and predictions

In the Qwak Inference lake, every inference is represented by a separate row. You can track the distribution, performance, and changes in your model behavior across times or builds.

Model Feedback

The Qwak Inference lake can collect the actual results of the model. After sending the actual results to Qwak Lake, we will automatically attach them to the relevant predictions data, allowing you to track your model performance metrics in real time.


Since you already have all the data you need in Qwak Inference Lake, you can use the Qwak Inference Lake data for retraining purposes, which can be fully automated with Qwak Automations.

Integrate with other tools

The Qwak Inference Lake data can be integrated with third-party vendors, from Audit tools to data and model monitoring tools, without any changes to your model code.

Get started today.

No commitments. No risk.