Repeat founder with a passion for computer vision. Currently building @finegrain_ai. Previously: built @moodstocks (acquired by Google).

Paris
Joined November 2007
A French engineer who lives quietly in Paris has spent 30 years writing software that the entire internet now runs on without knowing his name. He wrote the code that streams every YouTube video, every Netflix show, every TikTok clip. He wrote the code that runs the virtual servers underneath AWS, Google Cloud, and Microsoft Azure. He calculated more digits of pi than anyone in history. He has no Twitter. He has no marketing. He just keeps shipping. His name is Fabrice Bellard. Here is the story, because almost nobody outside the systems programming world knows what one man has built. Fabrice was born in 1972 in Grenoble, France. He studied at École Polytechnique, the top French engineering school. He never went to Silicon Valley. He never built a startup empire. He just wrote code. In 2000 he started a project called FFmpeg, an open-source multimedia framework for encoding, decoding, and streaming video. He was 28. The project did one thing nobody else had done well. It handled every video and audio format that existed, in one library, on every operating system. He led it himself for years. Today FFmpeg is the invisible engine of the internet. YouTube uses it. Netflix uses it. VLC uses it. Chrome and Firefox use parts of it. Every Android phone, every iPhone, every smart TV, every video editing tool you have ever touched runs FFmpeg somewhere underneath. If you have watched a video on a screen in the last 20 years, Fabrice's code processed it. He was not done. In 2003 he started QEMU, a machine emulator and virtualizer. He wrote it solo until version 0.7.1 in 2005. QEMU lets you run any operating system on any other operating system. It became the foundation of modern virtualization. KVM, the Linux kernel hypervisor, runs on top of QEMU. Every major cloud provider, AWS, Google Cloud, Microsoft Azure, IBM Cloud, runs virtual machines on infrastructure built around it. The Quick Emulator is the most cited piece of cloud infrastructure code on Earth. He kept going. In 2001 he won the International Obfuscated C Code Contest with a small C compiler that grew into TCC, the Tiny C Compiler. TCC can compile and boot a Linux kernel from source in under 15 seconds. In 2004 he calculated the most digits of pi ever computed at the time, using a personal desktop computer and an algorithm he derived himself called Bellard's formula. In 2011 he wrote a complete PC emulator in pure JavaScript that runs Linux in your browser, a project called JSLinux that engineers still cannot believe is real. In 2019 he released QuickJS, a small but complete JavaScript engine that fits where V8 cannot. In 2021 he released NNCP, a neural network based lossless data compressor that immediately took the lead on the Large Text Compression Benchmark. Then he turned his attention to large language models. He built TextSynth Server, a web server with a REST API for running LLMs locally. He released ts_zip and ts_sms, compression utilities that use language models to compress text and short messages at ratios traditional algorithms cannot reach. He released TSAC, a very low bitrate audio compression system. In December 2025 he released Micro QuickJS, a new JavaScript engine for microcontrollers, separate from QuickJS, designed for environments with almost no memory. Fabrice co-founded a telecom company called Amarisoft in 2012, where he serves as CTO. Amarisoft builds 4G and 5G base station software used by carriers and labs around the world. He has been running it for over a decade while continuing to ship personal projects from his own home page at bellard dot org He has no Twitter. He has no Instagram. He gives almost no interviews. His personal website is a flat list of projects with no styling, no fonts, no marketing copy. Just titles and links. A quiet French engineer who never moved to Silicon Valley wrote the code that quietly runs the internet. He is still shipping.
117
1,373
160
7,282
1,274,183
dfbrule retweeted
Had a company meeting this morning. Glad to see everyone showed up.
174
244
41
5,789
207,452
I wrote an interactive article explaining the geometric intuition behind Rectified Flows. I visually explain why flow-models tend to learn curved trajectories, why this is bad for sampling latency, and a relatively simple technique for mitigating it. Check it out! Link 👇
18
161
19
1,243
159,520
dfbrule retweeted
Artist Milly Bampini is so talented and skilled that she doesn't need Photoshop, only paint and brushes. [📹 millybampainti]
12
23
2
198
46,381
dfbrule retweeted
The great liberation of working on a niche product — whether that's Fizzy or Omarchy — is that you don't carry the burden of mass-market criticality or appeal. So you're free to have fun, be quirky, be whimsical.
23
29
6
635
70,675
Apple: "“After careful evaluation, we determined that Google’s technology provides the most capable foundation for Apple Foundation Models" theverge.com/news/860521/app…
568
831
302
12,294
1,084,799
Finegrain vs. ChatGPT: Which object eraser does it better? See for yourself!
1
4
313
dfbrule retweeted
Gradio
@Gradio20 Mar 2025
✨ Lights on, AI activated ✨ Upload ANY photo with lamp(s). Watch this app instantly illuminate it. No Photoshop. No editing skills needed. Just pure AI magic.🔥 Try Finegrain Light Switcher now.
13
77
6
379
63,017
Super COOL - You no longer need any paid app to do this. Finegrain Box Segmenter Model - High Definition (1024x1024) object-cutout model, making no assumption on what is background and what is foreground: users are fully in control. Published under the MIT license --- Existing off-the-shelf solutions, has some drawback: - Different people will have different definitions for background and foreground in a given image - a way to prompt these models is missing. - On the other hand, new promptable approaches like SAM or SAM2 don't meet the quality bar for the use cases we are pursuing: they are generating internally a 256x256 low resolution mask - with built-in upscaling mechanisms that create artefacts and struggle with complex masks (a la Eiffel Tower). The Finegrain Box Segmenter avoids these pitfalls by training MVANet
Gradio
@Gradio10 Sep 2024
Object Cutter Create high-quality HD background removal for ANY object in your image with a text prompt or bounding boxes!
4
46
1
348
44,156
How-to Magic Erase by Prompt, by @finegrain_ai It's free to try on Hugging Face: It works by both prompting and masking (see "bounding box") huggingface.co/spaces/finegr…
7
17
5,834
dfbrule retweeted
Gradio
@Gradio21 Aug 2024
Enhance your Images using AI! 🤩
3
23
2
158
12,627
There I said it, refiners abstractions are 10x @diffuserslib I wouldn’t be surprised if it takes off in 2024, as it is way better suited to build complex pipelines
I announced the launch of Refiners' bounty program a week ago, and I'm pleased to announce that: -> Already 1 bounty completed 🎉 -> There are 5 bounties currently being completed! 😎 finegrain.ai/bounties 👈 #AI #MachineLearning #Python #PyTorch
2
4
331
Wrote my first technical blog post about the deep learning framework I've been developing @finegrain_ai : blog.finegrain.ai/posts/simp… I learned AI and have always been very inspired by fast.ai project from @jeremyphoward (tell me what you think senpai)
3
7
802
We (Finegrain) shipped a new release of our machine learning framework, with an example about how to generate a slime plush for its logo 😂 github.com/finegrain-ai/refi…
2
6
18
5,021
3000th developer registered on the #Moodstocks platform, yahoo! ;) moodstocks.com/
2
4
Image recognition technology: the world as a shop window gu.com/p/39zeh/tw
Print-to-Mobile Sees Emergence of Image Recognition and Augmented Reality ...What They ThinkSeveral compelling... bit.ly/12Lt9Tw
dfbrule retweeted
La reconnaissance de packaging pour mettre produit à son panier virtuel. Encore mieux que la lecture de code barre. pocket.co/s9CiG
1
1
For people who ask us how fast #Moodstocks is, the answer is: instant. youtube.com/watch?v=3EqFNQxe…