Real time machine learning in Go
summary
Open Energi builds a virtual power station that learns each asset’s response and reports 2-second reporting window capacity so the fleet can be ramped up or down minute-by-minute; the system runs models written in Python but exposed as Go microservices via RPC, uses golearn and go ML for pure-Go experiments, persists model parameters, validates continuously with an error-priority queue, and claims it’s roughly 20x cheaper than building a physical power station for the same balancing service.
AI-written summary · about
chapters (17)
- 0:00Start
- 1:28Definitions of Machine Learning
- 3:15Supervised Learning
- 6:03Pre-Processing
- 6:58Learning Phase
- 8:08Validation
- 9:37Inputs
- 11:08Spam Detection
- 12:19Deep Minds Research
- 13:49Edison Power Station
- 15:10Virtual Power Station
- 17:57Load Behavior Forecasting
- 21:26Go Ml
- 23:06Bindings for Popular Frameworks
- 24:26Persisting the Model
- 25:40Data Scientist Is Not a Software Engineer
- 27:44Micro Service Wrapper
related talks
50mDeep Learning with Go
06mIntermittent Learning: On-Device Machine Learning on Intermittently Powered System
30mReal-Time ML services: From ideation to production
27mGreening the AI Cloud: Validating Power Models for Kubernetes Containers
20mBuilding GenAI That Doesn't Cost the Earth
53mProgramming Languages & the Power Grid
20mDecentralizing Compute for Machine Learning
33mspaGO: a work-in-progress yet ready-to-use library for ML/NLP