This is what the future of RTL development looks like to me https://t.co/WuV0x1DFCf
Thanks to @tlbflush agentic system -- this all was cooked semi-automatically (but now of course as a human I still have to eyeball it for high-order bit violations)
Fun project of the week: https://t.co/Z8moHaNVC3 where I create a git-friendly way to instruct and coordinate a pool of agents on long term jobs, with repository specific roles. As a success story, one of my projects (https://t.co/BOTF4oevTa) was completely cleaned up and fixed from a vibecoded mess autonomously, and the git commit shows that and the agent processes. I used https://t.co/lKvemBZy0L, because claude and codex were too cool to follow rules.
Every time we do #AIPlumbers events we make sure to film a couple of podcasts that translate the vibes of the conversations happening in the hallway track. Happy to have @ben_burtenshaw and @huggingface at FOSDEM Fringe 💥
new video on @huggingface youtube where I deep dive into how the kernels community makes it easier to build and use optimized kernels.
if you don't know, optimized kernels squeeze more out of hardware by adapting the operation with CUDA or C code.
they can be complex to build and distribute, sometimes requiring hours to install. the kernels community takes this down to a few seconds.
thanks to https://t.co/W2HSFZr002 for hosting the event and making the video.
Open-source AI stack startup @Ainekk0 and MRAM startup Veevx have merged! The combined company will gain Veevx's hardware team and its MRAM technology, already on the open-source tapeout Ainekko is working on:
https://t.co/JdirheO8Bb
We saved the best for last 😅 But really that seems to be a hot topic at the upcoming #FOSDEM, some community talks as well, i.e "Zero to matmul with the ET-SoC-1»! And some announcements on making ET platform even more open - from RTL to tape out!
Stay tuned!
Before @tlbflush updates us with new discoveries of working with open silicon platform (i.e. ET platform) at #FOSDEM2026#AIPlumbers Devroom, watch the first ever talk introducing it all!
https://t.co/90SE1QOPoa
#FOSDEM is almost here! #AIPlumbers#Devroom returns for the 2nd time on Sat, Jan 31. 80+ submissions this year — picking talks was painful😅Inference frameworks, model mgmt, compilers, quantization, open HW optimizations, schedule here: https://t.co/BBBHXBwMJ3
This #AIPlumbers talk showcased production-grade engineering: from offline performance configurators that find optimal cluster layouts, to dynamic K8s scheduling that understands physical GPU topology, coordinated multi-GPU serving, etc. Lot's of clever tricks 👀
Watch video from #AIPlumbers SF Edition presentation by @nvidia team on DYNAMO: Serving LLMs at AI-Factory Scale! And also really hope to hear more at #FOSDEM26 - don’t miss it!
https://t.co/Zaoj5tIrUa
🚨🚨🚨 There are only 5 DAYS LEFT TO APPLY TO SPEAK AT #FOSDEM26! If you have an #opensource project, doesn't matter if it's your hobby project or a company, and you want people to know about it. SUBMIT YOUR TALK here: https://t.co/iGiacEqBsk
Now that @Ainekk0 is officially out of stealth and more and more people are curious about our open source effort @AIFoundryorg (btw, welcome to our github and discord) it's time for proper introduction!
https://t.co/U68VmzwWHE
Many thanks to @sallywf for capturing our ethos and aspirations perfectly!
It truly takes a village to build the future of Open Source Edge AI compute platform.
Join us! Especially looking forward to @jzemlin and @linuxfoundation conversations to figure out the governance.
#FOSDEM26 is coming! Hurry up - submit your talk! Call for papers closes on Monday, 1 Dec 2025 CET (there will be no extensions!) Our DevRoom is called #AIPlumbers (surprise 😅)
Ainekko is hiring firmware developers, simulator experts, and programmers familiar with RISC-V and custom vector and tensor extensions for numerical algorithms.
What you’d be joining is an incredible hardware startup with a system view of what a compute platform should look like, from hardware design to the highest levels of the software stack.
Yes, we’re working on AI, and we’re open source.
The team is distributed, with headquarters in San Francisco.
If you’re interested, DM me.
#AIPlumbers SF edition by John Leimgruber also known as ubergarm is one of the most productive quantizers in AI world. I'm not joking - he has about 30TB of quants on @huggingface, they even started to limit his uploads (all fixed now, no worries 🤗) https://t.co/shX2sqLYmo
It's time for the videos from #AIPlumbers SF edition! @DhruvDiddi from Solo Tech walked us through the demands of #physicalAI from chips to the fine tuning of open robotics models. There are also some demos - so watch till the end!
https://t.co/QnWJXQTfQ2
The (unofficial 😉) #TechCrunch Hardware Afterparty at Studio 45 was packed! 300+ RSVPs, live demos, and nonstop energy from the Bay Area hardware scene 🦾
The true magic of #AIPlumbers always emerges during the unconference and most fun part as well - the breakout sessions are unstoppable! We ended up with a platform and a system working groups - I guess that’s how the industry operates, not just our company 🎯
Gianluca Guida, Head of Software, @Ainekk0 introducing ET platform or how to do build silicon co-evolving with SW for AI inference and beyond. Why ET? Well, T - is for Transputer. And also the maskot is cute!