Hacker News Reader

sphyrna-029on Ask HN: What have you built privately with AI that you'd like to share?

Been working on a self hosted discord alternative for the past 6 months. Comes with most features you would expect of chat software (and a few more). Testing native clients right now https://github.com/Chatter-Org/Chatter

sphyrna-029on Ask HN: What are you working on? (September 2026)

At the time of starting the project, stoat did not have screen share capabilities. This was a non starter for our community. Fluxer looked promising, but quickly became overwhelmed by the self hosting docs. Chatter is for small to medium size communities that don't want to spend more time maintaining their app than using it. With chatter you forward a few ports, fill out the compose file, and you're set. Chatter is also a PWA, no native client updates to worry about.

sphyrna-029on Ask HN: What are you working on? (September 2026)

Is this question an AI litmus test?

sphyrna-029on Ask HN: What are you working on? (September 2026)

Chatter, a self hosted almost feature complete alternative to Discord: https://github.com/Sphyrna-029/Chatter

sphyrna-029on Ask HN: What are you working on? (September 2026)

Working on Chatter https://github.com/Sphyrna-029/Chatter. A self hosted Discord alternative for small to medium sized communities. Live chat, voice, screen share, video call, customizable profiles, and comprehensive admin panel. Too many other features to list here. Check it out!

sphyrna-029on Ask HN: Who is using FPGA for ML inference?

For small models it is very do-able on entry level FPGA. With a PYNQ-Z2 at int8 you could potentially fit 1m params. It would be a fun exercise in writing a inference driver and implementing systolic arrays on a FPGA.