This is a submission for the Redis AI Challenge: Beyond the Cache.
What I Built
Sometimes cool things happen behind the frontend. This is a suite tool of scheduled tasks and API server.
The scheduled tasks update Github's latest public events to Redis streams every minute and compute Github contributors' score every hour.
The API server exposes the event stream and a list of top 20 highest scoring contributors. The client can consume it while making minimum number of requests to Redis server, thereby reducing latency.
Demo
Public API server endpoints for event stream and top scoring contributors.
Requesting event stream through endpoint /event-stream.
Request the top 20 highest scoring Github contributors through endpoint /active-contributors?since=today
How I Used Redis 8
The first job is to create a stream of most recent Github events. This is achieved through a scheduled task that fetches Github events from the official Github API. This API is not designed to serve real-time use cases as stated by Github. However, using the low-latency Redis stream and setting the task to run every minute creates a lightweight event log that act almost like real-time stream.
The second task is to rank users by scores since today. You are probably familiar with using Redis's sorted set for ranking. But how do you get the top scores within today?
Create a sorted set for each hour of today, starting from 0 hour to the last hour. For example, if the current time is 2am 2025-08-11, then you will have three sorted sets whose keys are 2025-08-11:00, 2025-08-11:01. Then call ZUNIONSTORE command on these sorted sets, with the destination being 2025-08-11:sum. The hourly data is aggregated from GhArchive so nothing slips through the cracks.
)
This feature is especially extensible thanks to how versatile ZRANGE is. You can add more query parameters to finetune this command to return X lowest scoring contributors or contributors who place 10th to 20th since this week, this month etc.
Top comments (0)