Skip to content

[Feature]: FLAME Integration with Kubeflow Katib #326

Open
@akramIOT

Description

@akramIOT

Is your feature request related to a problem? Please describe.

PROBLEM: As an AI_ML persona of end user using the flamectl commands I have issues in doing certain HP parameter tuning when i run the flib in an IOT Hardware platform IR1101 usung minikube.

IR1101 Platform spec: https://www.cisco.com/c/en/us/products/collateral/routers/1101-industrial-integrated-services-router/datasheet-c78-741709.html

SUGGESTION:

Since FLAME is supporting workloads to run in a de-coupled manner with the data plane workload running in edge nodes, its important to support Hyper Parameter tuning for DL in an optimum manner. In this specific context, is there a plan to support any API level integration with MLOps tools like 1) Kubeflow and 2) If support for Kubeflow is there/planned in roadmap then can we have an API from FLAME control plane/API server to tweak/parametrize Kubelfow Katib component for HP tuning to improve model efficacy ?

A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like
A: Optimum parameter tuning within a short time Convergence.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.
A: I know that FLAME leverages MLFLOW but any other API based solution integration to existing MLops tools like Kubeflow or Airflow or others ?

Additional context
Add any other context or screenshots about the feature request here.
A: To make the day to day routine work /life of an AI_ML engineer persona of end user easier when they use FLAME project.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions