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

  Log an issue

  Fork and edit

  Discuss on Discord

About this Learning Path

Skill level:  Introductory
Reading time:  30 min
Last updated:  11 Sep 2026
Skill level:
  Introductory
Reading time:
  30 min
Last updated:
  11 Sep 2026
Author:Pareena Verma, Arm
Arm IP:Neoverse
Tags: ML Google Axion Linux Flyte Python gRPC
Author:
  • Pareena Verma, Arm
Arm IP:
Neoverse
Tags:
ML Google Axion Linux Flyte Python gRPC

Who is this for?

This is an introductory topic for developers, data engineers, and ML engineers who want to build scalable machine learning workflow pipelines on Arm64-based Google Cloud C4A virtual machines (VMs) using Flyte workflow orchestration and gRPC-based microservices.

What will you learn?

Upon completion of this Learning Path, you will be able to:

  • Deploy Flyte workflow pipelines on Google Cloud C4A VMs powered by Axion processors.
  • Build distributed machine learning pipelines using Flyte tasks.
  • Implement gRPC-based services for feature engineering.
  • Integrate Flyte workflows with distributed services.
  • Run scalable ML pipelines on Arm-based cloud infrastructure.

Prerequisites

Before starting, you will need the following:

Summary

AI-assisted

This summary was drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
You’ll build a machine learning pipeline on Google Cloud C4A VMs using Flyte and a gRPC feature-engineering service. First, you’ll prepare an Arm64 environment, install the required components, and connect the service to Flyte tasks. The workflow loads and preprocesses data, generates features, trains a model, and evaluates your result on the Axion instance.

Frequently asked questions

AI-assisted

These FAQs were drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
Which Google Cloud instance type should I create?
Use the C4A instance family and select c4a-standard-4 (4 vCPUs, 16 GB memory).
Which operating system image should I choose for the VM?
Use a SUSE Linux Enterprise Server (SLES) arm64 image to prepare the development environment.
How do I know that the gRPC feature engineering service is integrated correctly with the workflow?
During execution, the workflow’s feature generation step calls the gRPC service and passes the resulting features to downstream tasks. If the service is unreachable or misconfigured, the dependent step won’t complete.
Is this environment single-node or multi-node, and where do components run?
The development environment uses a single-node setup on the Axion C4A VM. The gRPC feature engineering service runs as an external microservice that the Flyte workflow invokes.
How do I start the gRPC feature service before running the workflow?
Activate the flyte-env virtual environment and run python feature_server.py from the project directory. Leave that terminal running, then open a second terminal and run python workflow.py so that the Flyte workflow can connect to the service on port 50051.
Next