MLOPS retraining

[fawzi]

https://docs.databricks.com/en/machine-learning/mlops/mlops-workflow.html#7-retraining

In this article, it is mentioned that we can trigger retraining from the alerts.

• Triggered. If the monitoring pipeline can identify model performance issues and send alerts, it can also trigger retraining. For example, if the distribution of incoming data changes significantly or if the model performance degrades, automatic retraining and redeployment can boost model performance with minimal human intervention. This can be achieved through an SQL alert to check whether a metric is anomalous (for example, check drift or model quality against a threshold). The alert can be configured to use a webhook destination, which can subsequently trigger the training workflow.
  
  How to obtain the URI of the webhook of a specific job workflow?
  

[Kaniz_Fatma]

Hi @fawzi,  Let’s delve into the world of webhooks and discover how to obtain the URI for a specific job workflow.

Webhooks are a powerful mechanism for enabling real-time communication between applications. They allow apps to notify each other about specific events, which can then trigger actions. In your case, you’re interested in using a webhook to trigger retraining based on alerts.

Here’s how webhooks work:

1. Setting Up a Webhook:

   • Imagine you’re attending an event, but the exact date hasn’t been announced yet. Instead of repeatedly asking when it will happen, you put yourself on the waitlist. Once the date is set, you’ll receive the news.
   • Similarly, setting up a webhook involves configuring an app to listen for specific events. When that event occurs, the app sends an HTTP POST request to a designated URL (the webhook endpoint).

2. Webhook Workflow:

   • The process can be summarized in three steps:
     1. The webhook registers to listen for an event in System A (in your case, the monitoring pipeline or alert system).
     2. When the specified event occurs in System A, the webhook sends an HTTP POST request to a predefined URL (the webhook endpoint).
     3. System B (which could be your training workflow) receives the data from the POST request and takes action based on the event. This action is user-defined.

3. Obtaining the Webhook URI:

   • To obtain the URI (Uniform Resource Identifier) of a webhook for a specific job workflow, follow these steps:
     • Step 1: Identify the job workflow or process that you want to trigger via the webhook.
     • Step 2: Set up a webhook endpoint (a URL) that will receive the POST requests.
     • Step 3: Configure your monitoring pipeline or alert system to use this webhook endpoint as the destination.
     • Step 4: The webhook endpoint will now be the URI you need. It’s the URL where the alerts will be sent, and subsequently, it can trigger your training workflow.

4. Examples of Webhook Usage:

   • Webhooks are commonly used for real-time notifications, automation, and event-driven workflows. They eliminate the need for polling-based setups and allow immediate responses to events.

Remember that the webhook URI is specific to your setup, so ensure that you create a valid endpoint that can handle incoming POST requests. Once you’ve set it up, your alerts can seamlessly trigger the desired actions in your workflow! 🚀