Optimizing Cloud Costs with Machine Learning - A CloudVane Story
summary
A multi‑cloud management platform that forecasts daily and monthly costs, detects anomalies and recommends right‑sizing, claiming up to 40% savings on some workloads. Architecture uses Oracle for storage, a Python predictions API (based on FastAPI), Celery workers reading a Redis queue, and models chosen by data volume: simple averaging for single points, exponential smoothing for weeks, and gradient boosting with lag features for months. System auto‑selects strategy per resource, supports multi‑tenancy, backfills history and writes forecasts back to the DB.
AI-written summary · about
chapters (6)
- 0:00Start
- 2:36Challenges
- 6:46Need for Data Science
- 9:31Need for Machine Learning
- 16:20Machine Learning Flow Overview
- 17:40How to Scale Python
related talks
12mAI-Powered Predictive Analytics for Multi-Cloud Management
36mMigrating Applications to the Cloud in a Cost Effective Manner
21mBest Practices for Optimizing Your Cloud Spend with AI
42mHow We Saved Cloud Budget by Using Autoscaling
19mResource Management Innovation for AI Workloads in Cloud Computing
46mCloud FinOps: Empower real-time decision making
33mThe elephant in the cloud: Strategies of cloud cost optimisation
05mCutting 2/3rds of your cloud costs with common sense