Meetups/Infra/2025-02-03

(TODO summary)

Introductions

• [name] - [background]. [goals for meetup, or interests to explore]

• Loren - meetup -- host to talk
• Greg -- interested in self-hosted things generally, made an RSS reader in rust (with llms over the weekend)
• Oliver -- arrived today from Switzerland, travelling CA, bay today. Professionally a software engineer, home lab, cybersecurity & networking experience.
• Kat -- background in speech recognition, working for google for the last few years. Explore the open source domain again, rediscovering the work. Really interested in AI in the OSS space.
• Doug -- spending the last week chopping up images -- looking for a silver bullet -- difficult algorithms required. Write some software.
• Cloud -- used to write ... Got exposed to TTS/STT, Tor, encryption.

• Jake -- do backend dev at work, video games on the side, toying with llms lately, alwasy interested in self-hosted stuff
• Tucker -- first time here, note sure what's here. Video games, self driving cars, working with llms. Working on a project, self-hosted, jarvis -- extensive home automation
• Logan -- like playing minecraft, only work on, only play
• Moof -- here with Login, only play with rust, only thing I work. The game.

Lessons or Demo

• Read aloud: clarify for meetup. We are taking notes in a riseup pad (or I am--help appreciated, and links). We have meeting notes posted to the wiki. noisebridge.net, search Infra, or Meetups/Infra. (the Infrastructure page has a disambiguation link.)
• Shell, web services, self-hosting, networking!

Speech models, Nvidia -- ramping up, quietly but solidly. Reimplementing stuff published a few years ago.

Tokenizaion -- often happens in spectrum.

• https://rss.ichigo.everydayimshuflin.com/
• https://code.everydayimshuflin.com/greg/rss-reader

Written with cursor -- $20/mo. Have yet to try cursor against my own lengthy nix configuration.

Workflow with cursor, how fast was the iteration speed? It comple

• more introductions

JVM engineer.

• Theme: more -- what have you been building with llms --

Tucker -- jarvis home iOS app -- doing local development -- live update wihtout nginx, flask, llm

prompt -- you are jarvis -- you should act in this way.
a little bit of system context gets injected -- system status, e.g. speaker  disconnection, 
streaming back reply
11labs -- sentence by sentence. Lower latency but lower quality speech. 

bottlenecks

overnight sleep/dream/hallucination

secretary front end -- just dispatches task requests
task-specific requests
based on 

faiss -- 

phone / watch / vuez-its $800 (sp)/$500 -- (no camera -- 1) screen 2) custom code on it 3) look normal-ish -- would love mic/speaker/camera. Meta Orion project, $25k.

Last few things:

- aliexpress -- oura ring formfactor, can run python code, at least hr - streamlit -- snowflake open source -- ml demos, takes care of html - bunnylol -- / cony.

• Loren: make signs/designs, make networked LAN web app.

Questions, Discussion, or Coworking

• [Issue]

For next time

Readings & Exercises

• Readings
• Exercises

Join online

• Try it yourself!
  • Join libera.chat #nb-meetup-infra

https://www.noisebridge.net/wiki/Meetups/Infra