Introduction to MLOps at the Edge

on YouTube ↗47m

" summary

Claim: the cloud is "broken" for latency, cost, vendor lock-in, privacy and eight other reasons, so compute moves closer to sensors. Demonstrations use Barbara OS node manager with Docker workloads and a marketplace to deploy and batch-update models across thousands of devices, showing offline operation and local retraining for per-sensor models (Acciona water chemistry, EDP home energy). Measured pain: deployments often take two months to a year to reach production without an edge MLOps platform.

AI-written summary · about

" chapters (38)

1. 0:00intro
2. 0:22preface
3. 0:33the cloud is broken
4. 1:02¿why is it broken?...
5. 1:14dependence on the internet
6. 1:43high latency
7. 2:15service availability
8. 2:37loss of control
9. 3:19cost management
10. 4:14vendor lock-in
11. 5:04security and privacy
12. 5:43compliance
13. 6:25¿which is the solution? #edge is the new cloud
14. 7:04¿what is edge computing?
15. 8:54edge computing drivers...
16. 9:56lower bandwidth consumption
17. 10:48increased privacy and security
18. 11:22greater availability and autonomy
19. 12:13different types of edge
20. 15:07impact radar for 2023
21. 15:37index
22. 16:14acciona: intelligent virtual measurement of water components
23. 19:55edp: energy flexibility in self-consumption systems
24. 23:55edge computing ♥ ai - the architecture of alexa
25. 28:36maybe the cloud is not broken... maybe it just needs some friends
26. 28:56edge ♥ cloud ♥ ai - a powerful combination
27. 29:12¿how do we manage this complexity?
28. 29:43mlops - machine learning operations
29. 30:04edge ♥ cloud ♥ ai - mlops cycle
30. 30:52mlops challenges
31. 32:07architecture and frameworks...
32. 32:17edge computing platforms...
33. 34:43native providers
34. 35:23barbara
35. 35:33technological stack
36. 40:16demo
37. 45:32applications
38. 46:20thank you

" related talks

• 36mMLOps from the edge to on demand visualizationDavid Duncan · DevConf.CZ 2023
• 14mLinkEdge: Open-sourced MLOps Integration with IoT EdgeSavidu Dias · eSAAM 2023
• 05mThe Evolution of MLOpsAlex Yeh · Cloud Native + Kubernetes AI Day NA 2024
• 41mTHE STATE OF MLOPS - MACHINE LEARNING IN PRODUCTION AT ENTERPRISE SCALEBas Geerdink · Crunch Conference 2021
• 33mBuilding & Managing MLOps Pipelines in the CloudJoshua Arvin Lat · Global SRE Virtual Conference 2022
• 20mBringing CI/CD Practices to Machine Learning with MLOpsRobert Hafner · cdCon @ OSS NA 2024
• 08mMaximizing ROI Through Distributed Computing and Multi-CloudRavi Kumar Amaresam · Conf42 Machine Learning 2025
• 29mTeam Teardown: MLOps in the Cloud for IoTDanielle Dean, Mathew Salvaris, Mohan Mupiddi · TWIMLcon: AI Platforms 2021