tech / ml | empathy is the key for a safe AGI and human future MSc Data science - @unibocconi Ambassador @italianbldrs AI engineer for https://xitter.cf/t.co/HboEKGfw35
Italy
Joined September 2026
- Tweets140
- Following128
- Followers78
- Likes3.2K
"I do not love the bright sword for its sharpness, nor the arrow for its swiftness, nor the warrior for his glory.
I love only that which they defend"
- J.R.R. Tolkien's The Two Towers.
Here to remember how great friends Tolkien and Lewis were! The idea of one could be often found in the ideas of the other
About that, I also think that they will probably get heavily participated by the State. There is no way the private will stay in charge of such a key technology for the world.
"we can fix everything"
this is how you do it. delusional tech optimism is the way of the future. nor just in the industry, but in society as well
engineering = magic
I swear to God, if @dom_gag_96 creates a company to destroy the public sector and make it efficient with the use of AI agents, I'm in for help.
And a lot of us, Italian Builders, would be in it too.
I'm studying so much stochastic processes these days that I think I might as well start posting Markov chains and Martingales 😭
Giulio Sighieri retweeted
if an agent feels better than you at software, ask yourself what you were actually good at
if it was spinning up a monolith with a rest api, a database and a react frontend, then yeah, it will beat you at that every time. that was always the most replaceable part of the job
AI didn't make fundamentals less important, it made them the ONLY thing left. i’m talking about computer architecture, low level systems, data structures. that's what lets you look at what the model wrote and know WHY it's wrong
and we need to stop confusing the amount of working code AI produces with the ability to make good software. software has never been about who writes more lines faster
use it all day, i do. just don't build your whole skillset on the one layer it's best at 😂
You understand that the models we've now are very very good by noticing that the percentage of posts about them on tech X has been decreased drastically
Let me say: for young people (18-23yo) w/o deep software knowledge, the best thing to do is to keep studying and enlarge their knowledge base as much as possible.
Vertical agents, domain specific agents and all this things are for people already working in the industry. If all you do is vibecoding and following the trend, you will very probably not being able to create a real understanding of topics nor the capability of understanding the direction you want to pursue in your life.
So do a lot of projects, do startups, to stuff, but do not think "study is worthless": you can outsource the execution, not your understanding of the topic.
And someone who doesn't understand, is useless.
Now imagine: an alien student from a high-dimensional world built our reality as a simulation for his bachelor’s ML project. He mapped a small slice of his “brain” into a 3D creature, to see how much it could learn, from inside the simulation, about the reality underneath.
I would not be surprised: we will probably do the exact same thing.
Best italian community btw
Non so se lo sapete, ma gestisco una community privata su Telegram con oltre 800 builder italiani (@italianbldrs)
Voglio aprirla a nuove persone che costruiscono progetti, lavorano nel mondo tech/AI o semplicemente vogliono fare networking con persone che condividono i loro stessi interessi!
Se ti riconosci, commenta BUILDERS e ti mando l'invito in DM! 🚀
Now imagine opus 5.5, in two years from now, at that speed. Imagine it.
i’m today years old when i learned about chatjimmy.ai - you can just play with their demo right there
and holy cow i was NOT prepared for what 15000 tokens/s feels like. mind blown
apparently they achieved it by hardwiring model weights into the chip. and looks like the team got acquired by $AMD
i hope they keep at it and bring it to mainstream because this is probably one of the most magical things i experienced lately
We're getting so close it's crazy
🚀 Excited to introduce StreamRender-H3 — bringing real-time interactive worlds to life with diffusion models!
Built on Sol-Engine, we post-trained @MiniMax_AI H3 Ref2VA into a real-time streaming neural renderer.
🎮 Code drives the world. A game engine handles physics, object states, and player interactions, while H3 generates the visuals.
⚡ Real-time streaming: 2 denoising steps, 24 FPS, and ~273 ms per chunk on 8×GB200 GPUs.
🎨 One world, endless appearances: Change the reference image and prompt to render the same interactive world in entirely different styles.
🔥 Powered by streaming post-training, distillation, a fast causal H3 VAE, and full-stack inference acceleration.
Check out our demo and code! 👇
demo: nvlabs.github.io/Sana/Sol-En…
code: github.com/NVlabs/Sana/tree/…