Resource Monitoring and Optimization with Metaflow
summary
Metaflow integration adds a lightweight tracker that monitors CPU memory GPU VRAM network and disk at process and system level and claims a single line of code yields HTML reports plus instance recommendations. The project actually boots ~2500 server types and runs close to one million benchmarks to map real pricing and performance and the tracker is zero dependency on Linux. Demoed recommendations cut reserved memory from 10 GB to 1 GB saving about 90 percent and reports render after each step.
AI-written summary · about
chapters (6)
- 0:00Resource monitoring with Metaflow
- 4:14Spare Cores project overview
- 12:05Resource tracking package
- 20:11Metaflow and resource tracking
- 27:52Live demo and optimization
- 29:56Future plans & Q&A
related talks
40mOptimizing AI/ML Workloads
28mCompute anything with Metaflow
36mHuman-Friendly, Production-Ready Data Science Stack with Metaflow & Kubernetes
30mNavigating Cloud Expenses in Data & AI: Strategies for Scientists and Engineers
22mAI Observability At Meta Scale
44mHow Metaflow Became Netflix's Beloved Data Science Framework
32mHuman-Friendly, Production-Ready Data Science Stack With Metaflow & Kubernetes
35mBuilding Robust Reactive ML Systems In A Multiplayer Setting