Florian Dedov

🧠 AI/ML Engineer
📹 YouTube (485K+ Subs): GitHub:

[Florian Dedov @NeuralNine]

I was able to meet Jensen Huang the CEO of @nvidia at #GTCParis.

This guy is a legend!

[Florian Dedov @NeuralNine]

Looks like Kimi K3.1 Pro doesn't yap as much as other models.

That's great!

[Florian Dedov @NeuralNine]

I know it can be dangerous but I love the idea of running many agents in a multi-player environment. Unsupervised. It's quite exciting to see what happens.

[DHH @dhh]

I feel morally obligated to deliver as many antidotes to despair and cynicism as possible. The odds of p(bloom) are so overwhelmingly high. Can't lose good people to the black pill.

[Florian Dedov @NeuralNine]

@beffjezos Came in a bit late with my question but would appreciate an answer anyway: What are your thoughts on Europe's future in AI? And what would you do as an individual in Europe?

[Kimi.ai @Kimi_Moonshot]

Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.

🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.

Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on https://t.co/85jA71X9gw, Kimi Work, Kimi Code, and the Kimi API.

Open weights by Nov 4, 2026.

🔗 API: https://t.co/XCrgjXAqMw
🔗 Tech blog: https://t.co/5iPP0MBej6

[Florian Dedov @NeuralNine]

@beffjezos Probably every single piece of software has massive vulns hidden in plain sight. Now finding them requires no effort.

[Florian Dedov @NeuralNine]

@ClementDelangue These things look awesome. I would love to make videos about them on my channel. Sent you an e-mail in case you are open to it! :)

[Florian Dedov @NeuralNine]

Those open-source robots look sick!

Hey @pollenrobotics, any chance you'd send me one of each? Would love to make a YouTube video about them.

[Florian Dedov @NeuralNine]

@miaferrariii Actually have a setup like this in our startup.

Codex runs on a server. Claude uses an SSH key to connect to it and prompt it in headless mode.

[Florian Dedov @NeuralNine]

@PyTorch @linuxfoundation There's so much low-hanging fruit in small specialized models.

[Florian Dedov @NeuralNine]

I also think there is a lot of low-hanging fruit in small specialized models.

[PyTorch @PyTorch]

Smaller & specialized models are becoming dramatically more capable and new approaches to continual learning are enabling AI systems to improve from real-world experience after deployment. Jonathan Bryce, Executive Director of Cloud & Infrastructure at @linuxfoundation @CloudNativeFdn shares about how we're moving from bigger models to better systems during @AgenticAIFdn #AGNTCon + #MCPCon keynote.

The future of AI will depend not on any single model, but on an open set of technologies for building, optimizing, serving and continuously improving intelligence. The @PyTorch Foundation in bringing those technologies together, creating an open foundation that gives developers the freedom to build with any model, on any chip and across any cloud.

[Florian Dedov @NeuralNine]

@nvidia might be the most well-positioned company ever.

Disclaimer: I own $NVDA shares.

[Florian Dedov @NeuralNine]

I think secure software in the future will be proof-based.

[Florian Dedov @NeuralNine]

Retrospectively it's gonna be obvious to us that coding software is a task for machines, not humans.

[Pedro Domingos @pmddomingos]

AI is an intelligence lever. Meaning the smarter you are, the smarter it makes you.

[Santiago @svpino]

People are in denial.

They literally think they will wake up tomorrow and all the AI-generated code will somehow be useless, and we'll go back to typing code by hand like Neanderthals.

[Florian Dedov @NeuralNine]

@patrick_kuhnke_ would love to see how you're using it to come up with this hot take

[Florian Dedov @NeuralNine]

this is delusional

[Patrick Kuhnke @patrick_kuhnke_]

hot take: AI does not make you 10x faster.

[Nat Miletic @natmiletic]

People are insane if they think anyone is going back to manual coding and that we will somehow revert back to typing code by hand

[Florian Dedov @NeuralNine]

New Video: Databricks Crash Course (Sponsored)

We cover going from CSV files to deployed models. Spark Preprocessing, MLflow Tracking, Model Registry, Delta Tables & Model Serving.

A must-know if you want to go into MLOPs.

https://t.co/MYOibLsRM4