Build your own timeline algorithm
summary
A reproducible timeline stack claims to let anyone own and run ranking locally: it pulls from Mastodon via a Python API, uses LlamaFile + miniLM to compute embeddings, and exposes tools in the Marimo notebook UI to visualize embeddings, do semantic search and rerank timelines. Performance is practical — it encodes ~800 posts in 4 seconds and runs on a Raspberry Pi — so the prototype works on low‑power hardware, preserves privacy and is fully inspectable.
AI-written summary · about
chapters (7)
- 0:00Introduction and motivation
- 3:11What are timeline algorithms
- 5:59Problems with existing algorithms
- 7:04Introducing Bota project
- 8:08Tools behind Bota
- 10:36What you can do with Bota
- 13:45Performance and next steps
related talks
11mBuilding Real-Time Feature Pipelines
46mIncremental, zero-config Code Nav using stack graphs
01h 14mFuture of social media what’s the recipe for pluralism, long term user value ..
33mWhat If the Feed on Reddit Was More Than Just Posts
22mFrom 3000ms to 300ms: Lessons learned from optimizing Tanstack ...
29mBuilding a Time Machine with ES6 Proxies
24mPrickly by Nature: TanStack in the Wild
42mThe Jamstack can do that?!