Build ML Workflow Pipelines with Flyte and gRPC on Google Cloud C4A Axion processors
Introduction
Understand Flyte and gRPC ML workflows on Google Axion
Create a Google Axion C4A Arm virtual machine
Install Flyte and gRPC tools on Axion
Build a gRPC feature engineering service
Create ML Training Workflow
Execute and validate the ML pipeline
Understand the distributed ML architecture
Next Steps
Build ML Workflow Pipelines with Flyte and gRPC on Google Cloud C4A Axion processors
Introduction
Understand Flyte and gRPC ML workflows on Google Axion
Create a Google Axion C4A Arm virtual machine
Install Flyte and gRPC tools on Axion
Build a gRPC feature engineering service
Create ML Training Workflow
Execute and validate the ML pipeline
Understand the distributed ML architecture
Next Steps
Run the distributed ML workflow
In this section, you execute the distributed machine learning pipeline built using Flyte and gRPC.
The ML workflow will:
- load a dataset
- Preprocess the data
- generate features using a gRPC microservice
- train a model
- evaluate model performance
The feature engineering service runs independently and communicates with the workflow using gRPC remote procedure calls.
Start the feature engineering service
Make sure the flyte-env virtual environment is active. If you opened a new terminal, reactivate it:
source ~/flyte-env/bin/activate
Start the feature engineering service that was created in the previous section.
python feature_server.py
The output is similar to:
Feature gRPC service running on port 50051
Leave this terminal running because the ML pipeline will send requests to this service.
Run the ML workflow pipeline
Open a new terminal session. Navigate to the project directory.
cd ~/flyte-ml-pipeline
Run the workflow:
python workflow.py
Example pipeline execution output
The output is similar to:
Loading dataset
Preprocessing dataset: 10
Training model with feature: 200
Model accuracy: 10.0
Pipeline result: Model performance good
What happens during execution
During pipeline execution the following steps occur:
- The dataset is loaded by the Flyte task.
- The dataset is preprocessed.
- The workflow sends a request to the gRPC feature engineering service.
- The gRPC service generates features.
- The workflow uses the generated features to simulate model training.
- The model performance is evaluated.
- The pipeline returns the final result.
Pipeline execution flow
Load Dataset
│
▼
Preprocess Data
│
▼
Feature Engineering (gRPC Service)
│
▼
Model Training
│
▼
Model Evaluation
│
▼
Pipeline Result
Verify the gRPC service interaction
You can observe activity in the terminal running the feature service. When the workflow sends a request, the service prints a message similar to:
Feature gRPC service running on port 50051
Generating feature for: 20
The output confirms that the Flyte workflow successfully communicated with the gRPC service.
What you’ve learned and what’s next
In this section, you learned how to:
- Start the gRPC feature engineering service
- Execute the Flyte ML workflow pipeline
- Observe task execution across distributed services
- Verify communication between the workflow and the microservice
In the next section, you will explore the architecture of a distributed ML training pipeline implemented with Flyte and gRPC on Axion infrastructure.