AWS SageMaker
BetaThis source is currently in beta. It works end to end, but you may still hit rough edges. Let us know if you do.
Let AI connect your sources for you
Skip the manual setup — run this in your project and the wizard auto-detects your databases and APIs and connects them to PostHog.
Connect AWS SageMaker to PostHog to sync your data into the PostHog data warehouse for analysis and modeling.
Sync SageMaker resources from one AWS region. Grant these IAM permissions for the tables you select: sagemaker:ListTrainingJobs, sagemaker:DescribeTrainingJob, sagemaker:ListProcessingJobs, sagemaker:DescribeProcessingJob, sagemaker:ListEndpoints, sagemaker:DescribeEndpoint, sagemaker:ListModels, sagemaker:DescribeModel, sagemaker:ListEndpointConfigs, sagemaker:DescribeEndpointConfig, sagemaker:ListModelPackages, sagemaker:DescribeModelPackage, sagemaker:ListHyperParameterTuningJobs, sagemaker:DescribeHyperParameterTuningJob, sagemaker:ListAutoMLJobs, sagemaker:DescribeAutoMLJobV2, sagemaker:ListTransformJobs, sagemaker:DescribeTransformJob, sagemaker:ListPipelines, sagemaker:ListPipelineExecutions, sagemaker:DescribePipelineExecution, sagemaker:ListExperiments, sagemaker:DescribeExperiment, sagemaker:ListTrials, sagemaker:DescribeTrial, sagemaker:ListTrialComponents, sagemaker:DescribeTrialComponent, and sagemaker:ListMonitoringExecutions. Jobs, executions, experiments, endpoints, and model packages use full refresh to capture status changes. Models and endpoint configurations also support incremental sync by creation time.
Configuration
| Option | Description |
|---|---|
AWS access key IDType: text Required: True | |
AWS secret access keyType: password Required: True | |
AWS session tokenType: password Required: False | Temporary credentials require a session token. Scheduled syncs fail when the token expires. |
AWS regionType: text Required: True |
Linking AWS SageMaker to PostHog
- Go to the Data pipeline page in PostHog
- Click New source and select AWS SageMaker
- Fill in the required configuration fields
- Click Next, select the tables you want to sync, and then press Import
Supported tables
| Table | Description | Sync method | Incremental field | Primary key |
|---|---|---|---|---|
training_jobs | Training jobs with status, resource settings, metrics, and billable duration. | Full refresh | — | — |
processing_jobs | Processing jobs with status, input and output settings, and compute resources. | Full refresh | — | — |
endpoints | Inference endpoints with status, deployment settings, and production variants. | Full refresh | — | — |
models | Models with container settings, execution roles, and creation times. | Incremental, Full refresh | creation_time | — |
endpoint_configs | Endpoint configurations with production variants, instance types, and data capture settings. | Incremental, Full refresh | creation_time | — |
model_packages | Model packages and Model Registry versions with approval status, metrics, and inference settings. | Full refresh | — | — |
hyperparameter_tuning_jobs | Hyperparameter tuning jobs with objectives, best training jobs, and consumed resources. | Full refresh | — | — |
automl_jobs | AutoML jobs with problem type, objective, status, and best candidate. | Full refresh | — | — |
transform_jobs | Batch transform jobs with status, model, input and output settings, and run times. | Full refresh | — | — |
pipeline_executions | Pipeline executions with status, failure reason, and run times for every pipeline. | Full refresh | — | — |
experiments | Experiments with source, display name, and creation details. | Full refresh | — | — |
trials | Experiment trials with their experiment, source, and metadata. | Full refresh | — | — |
trial_components | Trial components with status, parameters, artifacts, and metric summaries. | Full refresh | — | — |
monitoring_executions | Model monitoring executions with status, schedule, endpoint, and failure reason. | Full refresh | — | — |