1 unstable release
0.1.0 | Mar 16, 2019 |
---|
#2 in #confusion
9KB
204 lines
Goals
- Drift monitoring
- Accuracy (and confusion matrix) monitoring
- Extensibility (custom functions for monitoring)
- Stateless
- Able to be used in AWS Lambda, Containers, etc
Open questions
- Choice of language (start with rust)
- Do we use Grafana or ELK or just use our own visualization?
- If we use our own visualization, what libraries?
Proposed architecture
- Thresholds set via yaml
- Simple accuracy (and other confusion matrix) metrics via yaml (timeframe to look back at, etc)
- Allow custom functions via CLI (which will generate the required code in the background with only the logic needed to be written)
- Custom functions will be pure functions
Dependencies
~0.6–0.9MB
~14K SLoC