Abstract:
As artificial intelligence rapidly reshapes software delivery, understanding the modern AI software stack is critical for engineers, architects, and DevOps professionals. This talk dives deep into the AI-powered software stack, covering how to design, build, and deploy scalable AI/ML workloads using cloud-native best practices.

Through real-world examples and a live demo, the session will explore how cutting-edge tools across the MLOps and DevSecOps lifecycle come together to deliver resilient, explainable, and automated AI systems at scale.

Key Topics Covered:

Overview of the modern AI software stack: From data ingestion to model serving

Integrating AI pipelines with Kubernetes, serverless platforms, and GitOps

Leveraging open-source tools like MLflow, Kubeflow, Ray, LangChain, and Hugging Face

Security and compliance in AI workloads (e.g., model explainability, SHAP, drift detection)

Observability, reproducibility, and automation in AI lifecycle

Live demo of an AI-powered cloud-native assistant ("CloudyBot")

Audience Takeaways:

A blueprint for building production-grade AI systems using cloud-native tools

Actionable practices for securing and monitoring ML workflows in real time

Insights into how organizations like PayPal are adopting AI/ML for scalable operations

How to integrate generative AI tools into enterprise DevOps ecosystems

Target Audience:
Software engineers, MLOps/DevOps practitioners, cloud architects, technical leaders, researchers, and AI/ML professionals