Davide Eynard (+mala)

Hi everyone! Updated #introduction after >3yrs 😱

It’s a me, +mala! 🇮🇹🇬🇧

Academic at heart, I love teaching. Been doing #OpenSourceAI at @MozillaAI@mastodon.social for the last 3yrs now. I believe ppl should own their AI stack.

Projects:

• 3564020356.org (>25yrs 😅)
• #PicoGopher
• BYOTA (Build Your Own Timeline Algorithm)
• current maintainer of https://github.com/mozilla-ai/llamafile🦙📁

Interests: #bouldering #gopher #SelfHosting #reversing #ML #solarpunk

Blog: https://aittalam.github.io gopher.3564020356.org

💙

Next week I’ll be in #Amsterdam for https://luma.com/rfujxz40

First things first: all my gratitude to @internetarchiveeurope@toot.community for hosting us 💙 I am already looking forward to meeting wonderful people and having interesting discussions!

Then, we’ll speak about something we have been working on for a while now: open source AI agents and tooling that makes them better for humans.

When I build projects that involve hardware, the outcome is usually ugly af (remember my “bomb edition” #PicoGopher?). I think the reason is me always going for the quickest and dirtiest solution until the result looks like a mix of the inventions of the protagonist’s dad in Gremlins and those by Data from the Goonies. This BTW is probably the worst-looking one (courtesy of “La focaccia del marinaio” which provided the plastic scaffolding)

While listening to Prokofiev's Romeo and Juliet, I remembered the first time I heard it:

https://invidious.nerdvpn.de/watch?v=pwTtxbe2lHs

🧟

#Amiga #music #zombies

Leaving Bruxelles after another #FOSDEM, I am grateful for the experience (standing at the #Mozilla booth with the @MozillaAI@mastodon.social t-shirt was a blast 🙂) and even more for the people I met: from Fediverse folks to ESC hackers, from our Mozilla siblings to mzai teammates I saw in person for the first time, from #AI luddites to curious ppl who coded agents with me on the fly, no encounter was any less than brilliant… And above all of this, no browser was harmed in the process 😬

After working forever in a branch parallel to main, I finally managed to ship a new llamafile release (0.10.0)! It’s been a fun quest and hopefully just the first of many.

An official post should be out soon, but I wanted to share this informally, as a personal life update: it’s been a few intense and cold months, with a lot happening at work, in the world, and in my personal life. And while not everything’s ok, still, another spring is here and I am grateful for it!

@b0rk@social.jvns.ca

(please don't tell me that you can ask generative LLMs to do searches)

I won’t tell you but… now I am curious about why 🙃 Is it because you already do it? Or because you don’t want to use them this way? And if you don’t want to, why is that?

I wouldn’t ask Claude. But I like using open weights models to help me find material, links between things, and provide me with URLs and whatever extra info I need. And I found a combo of pi, local models and SearXNG to be pretty good at this.