Self-healing on existing ML pipeline

[sureshtmca]

I have existing ML pipeline built pre-era of gen ai. now can i integrate this assistant to this ML pipeline where some regression model has been deployed and working for a decade already. can this assistant monitor this pipeline and take decisions on retraining this current regression model automatically? if Yes, HOW. as i know, this assistant is used to create new ML model but integration with current ML pipeline and further self-healing aspects are looking fresh and not documented yet i guess

[FANGAreNotGnu]

Hi @sureshtmca, thanks for your interest in our work. AutoGluon Assistant (MLZero) is currently designed for creating new ML solutions from given data, rather than monitoring or managing existing production pipelines. For your use case, you could consider using MLZero to generate a newer solution on your data to potentially replace the existing algorithm. However, integration with your pipeline and the retraining logic would heavily depend on your specific infrastructure and product requirements, which is something you would need to build externally around the solution generated by MLZero.