Pipeline
Q. What is the Pipeline module?
A. The Pipeline module is a core feature of the MlAngles MLOps Platform that allows users to build and manage end-to-end machine learning pipelines. Each pipeline represents a configurable, repeatable workflow covering stages such as data analysis, ingestion, and preprocessing.
Q. How do I access the Pipeline module?
A. You can access the Pipeline module by selecting the Pipeline icon from the left-hand sidebar. This will take you to the module’s landing page.
Q. What is displayed on the Pipeline module landing page?
A. The landing page presents a tabular view of all existing pipelines. Each row in the table contains important metadata to help you identify and manage pipelines.
Q. How do I start creating a new pipeline?
A. To create a new pipeline, click the “New Pipeline” button located at the top-right corner of the Pipeline module page.
Q. Are pipelines reusable?
A. Yes. Pipelines are designed to be configurable and repeatable, making them reusable for similar ML workflows across different datasets.
Q. Does the platform help generate pipeline steps automatically?
A. Yes. Once you provide the Problem Statement, the platform’s integrated LLM (Large Language Model) suggests a set of pipeline steps. Users can review these steps and optionally remove or adjust them before saving.
Q. Who is responsible for configuring data source keys and repository credentials?
A. These must be preconfigured and managed by the administrator in the Access and User Management section and assigned to individual users.
Q. How does the Pipeline Code Generation Interface work?
A. This interface provides a visual flow diagram of pipeline steps and a prompt-based code generation panel. Users can:
- Reorder or delete steps in the flow
- Generate or customize code using natural language prompts
- Upload their own code files
- Save and run individual steps to test them
- Visualize charts or outputs from data steps
Q. How do I execute a single pipeline step?
A. Select the step, generate code, and click Save and Run. The output is streamed in the right-hand panel.
Q. How is data transferred between steps?
A. Data is transferred in .csv format. Ensure that after transformations, the dataset is saved as data.csv for use in the next step.
Q. How do I execute the entire pipeline?
A. The steps are
- Save the pipeline using the Save Pipeline button.
- The code is pushed to a repository.
- From the pipeline home screen, click Execute Pipeline.
- The platform will run all steps sequentially and redirect you to the Pipeline Execution Screen for monitoring.
Q. How do I monitor pipeline execution?
A. Each step shows one of the following statuses in real time:
- RUNNING – Step is currently executing
- FAILED – Step execution failed
- SUCCESS – Step completed successfully
Execution time is displayed alongside the status.
Q. Can I edit an existing pipeline?
A. Yes. Click Edit Pipeline on the pipeline home screen. After making changes, re-run and test all steps, then save the pipeline again.