

Kubernetes, AI Gateways, and the Future of MLOps // Alexa Griffith // #294
12 snips Mar 7, 2025
Alexa Griffith, a Senior Software Engineer at Bloomberg, shares her journey in building scalable ML inference platforms and contributing to open-source projects. She discusses the evolution of workflow tools, comparingAirflow with new solutions like CoopFlow and Argo. Highlighting the Envoy AI Gateway, she addresses the challenges of managing AI traffic. Alexa also emphasizes the importance of aligning tech work with business goals, optimizing GPU utilization, and fostering effective communication between teams for successful AI deployments.
AI Snips
Chapters
Transcript
Episode notes
Airflow Logs
- Alexa Griffith's first tech task involved improving Airflow logs.
- Initially, logs went into a database, and she pulled them for display during job runs.
Airflow's Past Limitations
- Four years ago, Airflow was the primary workflow tool despite its limitations.
- This spurred innovation, leading to the creation of new companies and tools in the pipeline space.
Argo Workflows Experience
- Alexa Griffith hasn't extensively used Argo Workflows.
- A team at Bloomberg manages it for certain jobs, finding it user-friendly.