AWS SageMaker

Connect your AWS SageMaker-powered application to Latitude Telemetry for feature-level observability and evaluations.

This integration is only available in the Python SDK.

Overview

This guide shows you how to integrate Latitude Telemetry into an existing application that uses AWS SageMaker for model inference.

After completing these steps:

• Every SageMaker call (e.g. invoke_endpoint) can be captured as a log in Latitude.
• Logs are grouped under a prompt, identified by a path, inside a Latitude project.
• You can inspect inputs/outputs, measure latency, and debug SageMaker-powered features from the Latitude dashboard.

You’ll keep calling SageMaker exactly as you do today — Telemetry simply observes and enriches those calls.

---

Requirements

Before you start, make sure you have:

• A Latitude account and API key
• A Latitude project ID
• A Python-based project that uses boto3 for SageMaker
• Configured AWS credentials

That’s it — prompts do not need to be created ahead of time.

---

Steps

1. Install requirements

   Add the Latitude Telemetry package to your project:

   pip

   pip install latitude-telemetry
   
2. Initialize Latitude Telemetry

   Create a single Telemetry instance when your app starts.

   You must enable the Sagemaker instrumentor so Telemetry can trace it.

   telemetry.py

   import os
   from latitude_telemetry import Telemetry, Instrumentors, TelemetryOptions
   
   telemetry = Telemetry(
       os.environ["LATITUDE_API_KEY"],
       TelemetryOptions(
           instrumentors=[Instrumentors.Sagemaker],  # This enables automatic tracing for SageMaker
       ),
   )
   

   The Telemetry instance should only be created once. Any SageMaker client calls made after this will be automatically traced.

3. Wrap your SageMaker-powered feature

   Wrap the code that calls SageMaker using telemetry.capture.

   You can use the capture method as a decorator (recommended) or as a context manager:

   Using decorator (recommended)

   import boto3
   import json
   from telemetry import telemetry
   
   @telemetry.capture(
       project_id=123,  # The ID of your project in Latitude
       path="generate-support-reply",  # Add a path to identify this prompt in Latitude
   )
   def generate_support_reply(input: str) -> str:
       # Your regular LLM-powered feature code here
       client = boto3.client("sagemaker-runtime", region_name="us-east-1")
   
       response = client.invoke_endpoint(
           EndpointName="your-llm-endpoint",
           ContentType="application/json",
           Body=json.dumps({"inputs": input}),
       )
   
       # You can return anything you want — the value is passed through unchanged
       result = json.loads(response["Body"].read().decode())
       return result["generated_text"]
   

   Using context manager

   import boto3
   import json
   from telemetry import telemetry
   
   def generate_support_reply(input: str) -> str:
       with telemetry.capture(
           project_id=123,  # The ID of your project in Latitude
           path="generate-support-reply",  # Add a path to identify this prompt in Latitude
       ):
           # Your regular LLM-powered feature code here
           client = boto3.client("sagemaker-runtime", region_name="us-east-1")
   
           response = client.invoke_endpoint(
               EndpointName="your-llm-endpoint",
               ContentType="application/json",
               Body=json.dumps({"inputs": input}),
           )
   
           # You can return anything you want — the value is passed through unchanged
           result = json.loads(response["Body"].read().decode())
           return result["generated_text"]
   

   The path:

   • Identifies the prompt in Latitude
   • Can be new or existing
   • Should not contain spaces or special characters (use letters, numbers, - _ / .)

---

Seeing your logs in Latitude

Once your feature is wrapped, logs will appear automatically.

1. Open the prompt in your Latitude dashboard (identified by path)
2. Go to the Traces section
3. Each execution will show:
   • Input and output messages
   • Model and token usage
   • Latency and errors
   • One trace per feature invocation

Each SageMaker call appears as a child span under the captured prompt execution, giving you a full, end-to-end view of what happened.

---

That’s it

No changes to your SageMaker calls, no special return values, and no extra plumbing — just wrap the feature you want to observe.