Panel: Argo for ML: Achieving Scalability and User Experience
summary
Companies describe building Argo Workflows platforms plus a Python layer that actually gets used: Argo plus Hera Python SDK powers continuous training and deployment across 300 engineers and over 40 teams, handling spiky cron-driven prediction deadlines and large fan-outs. Key wins are artifact visualization, intermediate parameters and provenance export; key pain is developer UX—YAML vs Python and onboarding—visible in an active support channel (5–10 daily questions). Experimentation with Argo Events and metrics/monitoring is underway.
AI-written summary · about
chapters (14)
- 0:00Panel introduction
- 1:27ML orchestration challenges
- 2:58Trading deadlines and ML
- 4:02Bloomberg chose Argo Workflows
- 6:31Centrica’s Argo adoption
- 7:58Hera and Python role
- 9:02Impact of Hera at Bloomberg
- 11:30Hera’s wildfire adoption
- 12:35Adoption challenges
- 17:10Future of ML with Argo
- 21:28Q&A
- 22:10Resistance to Hera
- 23:14Hera’s expansion to projects
- 24:20Closing remarks
related talks
10mScaling Time Series Analysis With Argo Workflows: Patterns and Practi...
27mEmpowering Autonomy: BYD's Journey Taming Million-Task Scale With Argo...
25mLeveraging Argo Workflows and Argo Events for Cloud Agnostic Enter...
07mArgo in an AI-Native World: Effective In-Cluster Machine Learning and A...
20mScaling Gene Therapy Research with Argo Workflows and Hera
23mPerception Data Pipeline Reconstruction With Argo Workflows
15mMaking Complex R Forecast App...
27mManaging Thousands of Automatic Machine Learning Experiments with...
