https://github.com/heardacat/seek-ml
Online Feature Store in Python, using Voyager
https://github.com/heardacat/seek-ml
Last synced: 2 months ago
JSON representation
Online Feature Store in Python, using Voyager
- Host: GitHub
- URL: https://github.com/heardacat/seek-ml
- Owner: HeardACat
- Created: 2023-11-05T02:44:47.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-11-05T02:46:45.000Z (over 1 year ago)
- Last Synced: 2025-03-26T16:40:09.718Z (2 months ago)
- Language: Python
- Size: 25.4 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Seek ML
An online feature store serving pattern that uses `voyager` and is completely in-memory.
Usage:
```py
from seek_ml.seek_store import SeekStore
store = SeekStore()
group: str = "group"
store.add(group, 2, np.arange(10))
store.add(group, 200, np.arange(10) + 1)
store.fetch(group, 2)
```
Since `voyage` provides both a simple key-value lookup and a vector store, we can provide a
simple abstraction to provide an online feature store. The rough idea is that at any point in time:
* push features to a particular group by entity id via `store.add(group, id, vector)`
* retrieve features by group and entity id via `store.fetch(group, id)`
This can help provide ways to access features in a multi-entity setup or in a setup where
each event generates different set of features.
This repository is also a simple exemplar repository with black, isort, ruff and pyright which
I want to iterate and use as a template.
**Dev Scripts**
```
just format # format
just lint # linting and static analysis
just test # running tests
```