Without being glib, I honestly wonder if Fabrice Bellard has started using any LLM coding tools. If he could be even more productive, that would be scary!
I doubt he is ideologically opposed to them, given his work on LLM compression [1]
He codes mostly in C, which I'm sure is mostly "memorized". i.e. if you have been programming in C for a few decades, you almost certainly have a deep bench of your own code that you routinely go back to / copy and modify
In most cases, I don't see an LLM helping there. It could be "out of distribution", similar to what Karpathy said about writing his end-to-end pedagogical LLM chatbot
---
Now that I think of it, Bellard would probably train his own LLM on his own code! The rest of the world's code might not help that much :-)
He has all the knowledge to do that ... I could see that becoming a paid closed-source project, like some of his other ones [2]
This is a funny one because on the one hand the answer is obviously no, it's very fiddly stuff that requires a lot of umming and ahhing, but then weirdly they can be absurdly good in these kinds of highly technical domains precisely because they are often simple enough to pose to the LLM that any help it can give is actually applicable immediately whereas in a comparatively boring/trivial enterprise application there is a vast amount of external context to grapple with.
From my experience, it's just good enough to give you a code overview of a codebase you don't know and give you enough implementation suggests to work from there.
In 2025, there is no shame in using an LLM. For example, he might use it to get help debugging, or ask if a block of code can be written more clearly or efficiently.
> Without being glib, I honestly wonder if Fabrice Bellard has started using any LLM coding tools
I doubt it. I follow him and look at the code he writes and it's well thought out and organized. It's the exact opposite of AI slop I see everywhere.
> He codes mostly in C, which I'm sure is mostly "memorized". i.e. if you have been programming in C for a few decades,
C I think he memorized a long time ago. It's more like he keeps the whole structure and setup of the program (the context) in his head and is able to "see it" all and operate on it. He is so good that people are insinuating he is actually "multiple people" or he uses an LLM and so on. I imagine he is quite amused reading those comments.
> I honestly wonder if Fabrice Bellard has started using any LLM coding tools. If he could be even more productive, that would be scary!
Thatâs kind of a weird speculation to make about creative people and their processes.
If Caravaggio had had a computer with Photoshop, if Eintein had had a computer with Matlab, would they have been more productive? Is it a question that even makes sense?
Itâs far from being impossible, the main thing you need is free time and obsession (and money for your free time btw).
C or asm are not obscure languages or anything, they are brutal languages where you have to trace runtime from A to Z, and manage the memory.
In 1990, it was absolutely normal to code in C. Yes you had to decode images yourself, yes you had to decode audio, yes you had to raytrace, etc.
âWait, you had to calculate all of these by hand ?
Yes my friend everybody had to do that in my time, what else could we do ?
So we took books, and did one by one.
This was the norm, just that it became some sort of archeology.â
Every year, thousands of 19-year-olds complete these tasks in low-level schools like Epita/Epita/42 or in demoscene contests. They aren't geniuses; they are just students who were forced to read the manual and understand how the computer actually works.
Free time wonât guarantee you success, but free time + obsession will (like Terry Davis).
Really, this is not alien tech.
Before FFmpeg, people had to encode the videos. Before emulators someone had to create the state machine, etc. All these people it would be insane to ignore them.
Most of the difficult problems have shifted somewhere else from low-level.
How to simulate millions of pharmaceutical molecules in short amount of time ?
How to simulate the world in GTA VI ?
Saving 2 bytes of memory by writing asm (that⌠wonât be portable) is not the thing going to save you. The problems are now elsewhere.
The problem now is not about âwow you read ancient manuals and mixed sand with water and got a solid foundational brickâ but it is about âok, using these bricks, how to build a skyscraper that is 1km tallâ.
No doubt that these modern programmers are as good as the archeologists who like to explore handcrafted code.
By the way (I'm who submitted the link) - I found a prior discussion on the same document but different URL now expired. There's even a comment by one of the authors, sharing some context.
While the guy is brilliant, I doubt he could fit the role of senior/staff/principal engineer in any one-level-below faang kind of company. Typically, these roles require good communication skills and working together with other engineers (which is really hard). So, while he's very good at the tech level, I think he primarily works alone? In that regard, it would be a very bad fit. I may be wrong, tho.
In technically deep domains like Bellard works in, Staff+ roles bias more towards technical expertise, and managers also tend to be more technical and able to more completely address technical coordination tasks. Sometimes we like to assume that if someone is good at one thing, theyâll be bad at something more mundane (to make ourselves feel better), but I sincerely doubt he would have any trouble in such a role.
Fabrice Bellard is not a 10x engineer, he is a 100x engineer. You could attach him to a good people manager and either build a team around him or allow him to work independently on a project that he finds exciting that also aligns with company goals.
I think you are mixing up art, technical skills and productivity.
Put Terry Davis (again him) as senior manager at Apple, and see the result.
From my point of view, Terry has the same level and approaches as Fabrice.
It does not guarantee at all that he is going to be more productive than 100 engineers as you directly claim.
It makes them good in what they like to do (writing obfuscated or low-level code, or implementing from scratch from specifications) as art or creativity.
Thank you for introducing me to Terry Davis. I'm going to read more about him.
I am definitely not talking about art.
When I refer to 100x engineer, I'm referring to the impact that QEMU and FFmpeg have had on the world. I would be surprised if anyone who is familiar with these two projects would disagree that they have been highly impactful.
At M.Bellardâs level one would could hardly even call such an outcome a character flaw, but my occasional privilege of managing - one should rather say, enabling - high performance teams, taught that the Venn intersection of âcompetent with imaginationâ and âcollegiate mannerâ is far from empty, even in the tech sector.
ââWe're delighted to have you here,â he said, âbut a word of advice. Don't try to be clever. We're all clever here. Only try to be kind, a little kind.â Like most university stories, this one is variously attributed and it probably never even happened but, as the Italians say, se non e vero, e ben trovato - even if it isn't true, it's well founded.â ⸺ Stephen Fry.
> While the guy is brilliant, I doubt he could fit the role of senior/staff/principal engineer in any one-level-below faang kind of company.
Maybe but whatâs the point? Hell, I might guess he is terrible at jiggling and basket weaving, too. Complete failure as wrestler, even. But that is kind of neither here or there. Or is it you think staff title at faangs is some kind of pinnacle position every engineer should strive for? It actually always strikes me as a funny title. In college when they didnât have a specific professor to teach or just going to use a grad student they put âstaffâ in the name box so in my mind itâs associated with a random lower rung student who couldnât get away doing just research.
Yeah, staff engineer is a pinnacle "still doing engineering and maybe leadership but not management" position in engineering firms. The academic "staff" is just a "not really one of us" gatekeeping-the-servants title.
The fact that so many people use FFmpeg and QEMU suggest that he is quite good at documenting, collaborating, and at least making his code remarkably clean and easy to follow. This already puts him way ahead of the average silicon valley senior software engineer that I've worked with. However, he does value independence so I don't think he would have been happy working at a faang-type company for long.
>Fabrice won International Obfuscated C Code Contest three times and you need a certain mindset to create code like thatâwhich creeps into your other work. So despite his implementation of FFmpeg was fast-working, it was not very nice to debug or refactor, especially if youâre not Fabrice
He might as well be but why would he give a flying fuck about it? He gets to do what he wants and is financially independent for doing just that. Most can only dream about it.
Myself - I do not come within a million miles to his professional level, but I still have managed to do just that - I develop what I want, how I want and get paid for it. I am 64 and still design and develop actively for my own company and for clients. Gives me happiness, motivation to stay alert and more than enough time to still do my hobbies (mostly various outdoor activities).
Publishing ffmpeg and QEMU in a five year span that also included winning IOCCC (twice!) is absolutely bonkers.
Without being glib, I honestly wonder if Fabrice Bellard has started using any LLM coding tools. If he could be even more productive, that would be scary!
I doubt he is ideologically opposed to them, given his work on LLM compression [1]
He codes mostly in C, which I'm sure is mostly "memorized". i.e. if you have been programming in C for a few decades, you almost certainly have a deep bench of your own code that you routinely go back to / copy and modify
In most cases, I don't see an LLM helping there. It could be "out of distribution", similar to what Karpathy said about writing his end-to-end pedagogical LLM chatbot
---
Now that I think of it, Bellard would probably train his own LLM on his own code! The rest of the world's code might not help that much :-)
He has all the knowledge to do that ... I could see that becoming a paid closed-source project, like some of his other ones [2]
[1] e.g. https://bellard.org/ts_zip/
[2] https://bellard.org/lte/
What I wonder is: are current LLMs even good for the type of work he does: novel, low-level, extremely performant
This is a funny one because on the one hand the answer is obviously no, it's very fiddly stuff that requires a lot of umming and ahhing, but then weirdly they can be absurdly good in these kinds of highly technical domains precisely because they are often simple enough to pose to the LLM that any help it can give is actually applicable immediately whereas in a comparatively boring/trivial enterprise application there is a vast amount of external context to grapple with.
As a professional C programmer, the answer seems to be no; they are not good enough.
From my experience, it's just good enough to give you a code overview of a codebase you don't know and give you enough implementation suggests to work from there.
If Fabrice explained what he wanted, I expect the LLM would respond in kind.
If Fabrice explained what he wanted the LLM would say it's not possible.
When the coding assistant LLMs load for a while it's because they are sending Fabrice an email and he corrects it and replies synchronously.
No
I doubt it, although LLMs seem to do well on low-level (ASM level instructions).
He has in fact written one: https://bellard.org/ts_server/
Yeah I've seen that, but it looks like the inference-side only?
Maybe that is a hint that he does use off-the-shelf models as a coding aid?
There may be no need to train your own, on your own code, but it's fun to think about
In 2025, there is no shame in using an LLM. For example, he might use it to get help debugging, or ask if a block of code can be written more clearly or efficiently.
> Without being glib, I honestly wonder if Fabrice Bellard has started using any LLM coding tools
I doubt it. I follow him and look at the code he writes and it's well thought out and organized. It's the exact opposite of AI slop I see everywhere.
> He codes mostly in C, which I'm sure is mostly "memorized". i.e. if you have been programming in C for a few decades,
C I think he memorized a long time ago. It's more like he keeps the whole structure and setup of the program (the context) in his head and is able to "see it" all and operate on it. He is so good that people are insinuating he is actually "multiple people" or he uses an LLM and so on. I imagine he is quite amused reading those comments.
Still, humans can only type so quickly. It's not hard to imagine how even a flawless coder could benefit from an llm.
Some talented people (mitsuhiko, Evan you) seem to leverage LLM their own way. Probably as legwork mostly.
I think it's the opposite: llms ask Fabrice Bellard instead
Congrats, the Chuck Norris meme has finally made its way onto HN.
Fabrice Bellard is far more deserving of the honor that olâ Chucky.
Tough choice: Knuth, Bellard, Norvig...
They're trained on his code for sure. Every time I ask about ffmpeg internals, I know it's Fabrice's training data.
Keep in mind even if someone writes their own code LLM is great to accelerate: tests, makefiles, docs, etc.
Or it can review for any subtle bugs too. :)
Why every single post in HN has to come down to talk about AI sloop...
Is Fabrice like the Chuck Norris of programming?
Hopefully without the politicsâŚ
> I honestly wonder if Fabrice Bellard has started using any LLM coding tools. If he could be even more productive, that would be scary!
Thatâs kind of a weird speculation to make about creative people and their processes.
If Caravaggio had had a computer with Photoshop, if Eintein had had a computer with Matlab, would they have been more productive? Is it a question that even makes sense?
> Is it a question that even makes sense?
Absolutely. It's a very intriguing thought invoking the opposite of the point you're trying to make.
Maybe today Bellard uses LLMs though
Matlab has been proven to be a indispensable tool in many fields.
AI is the same, for example creating slop or virtual girlfriends.
This biography includes more information than I've seen elsewhere about the legendary programmer, who's been discussed time and again on this forum.
He did a few things since, notably 5G base stations using PC hardware, and some LLM stuff.
And he wrote a proprietary ASN.1 compiler and stack.
Itâs far from being impossible, the main thing you need is free time and obsession (and money for your free time btw).
C or asm are not obscure languages or anything, they are brutal languages where you have to trace runtime from A to Z, and manage the memory.
In 1990, it was absolutely normal to code in C. Yes you had to decode images yourself, yes you had to decode audio, yes you had to raytrace, etc.
âWait, you had to calculate all of these by hand ?
Yes my friend everybody had to do that in my time, what else could we do ?
So we took books, and did one by one.
This was the norm, just that it became some sort of archeology.â
Every year, thousands of 19-year-olds complete these tasks in low-level schools like Epita/Epita/42 or in demoscene contests. They aren't geniuses; they are just students who were forced to read the manual and understand how the computer actually works.
Free time wonât guarantee you success, but free time + obsession will (like Terry Davis).
Really, this is not alien tech.
Before FFmpeg, people had to encode the videos. Before emulators someone had to create the state machine, etc. All these people it would be insane to ignore them.
Most of the difficult problems have shifted somewhere else from low-level.
How to simulate millions of pharmaceutical molecules in short amount of time ?
How to simulate the world in GTA VI ?
Saving 2 bytes of memory by writing asm (that⌠wonât be portable) is not the thing going to save you. The problems are now elsewhere.
The problem now is not about âwow you read ancient manuals and mixed sand with water and got a solid foundational brickâ but it is about âok, using these bricks, how to build a skyscraper that is 1km tallâ.
No doubt that these modern programmers are as good as the archeologists who like to explore handcrafted code.
Victor Taelin posts an intuition `HVM is missing a fundamental building block' having done 10 years thinking
> Itâs far from being impossible, the main thing you need is free time and obsession (and money for your free time btw).
I'm aware :(
(I maintain one, one written by my Swedish friends, whom too were obsessed.)
With his recent release of MicroQuickJS, and also prior work, he kind of has to do epic things. People expect that of him.
Do we think Bellard got rich, like antirez?
(2009)
I've put that in the title above, although the URL says 2020 - could the text have been updated?
By the way (I'm who submitted the link) - I found a prior discussion on the same document but different URL now expired. There's even a comment by one of the authors, sharing some context.
https://news.ycombinator.com/item?id=2555654 (2011)
My favorite line from the biography:
> [As a child] Bellard was drawn to electronic devices. His first word was magnĂŠtophone (tape recorder).
Nice find! Macroexpanded:
Fabrice Bellard [pdf] - https://news.ycombinator.com/item?id=2555654 - May 2011 (29 comments)
I guess we can throw these in too although it was presumably a different article:
Fabrice Bellard: Portrait of a Super-Productive Programmer (2011) - https://news.ycombinator.com/item?id=32795067 - Sept 2022 (26 comments)
Fabrice Bellard: Portrait of a Super-Productive Programmer (2011) - https://news.ycombinator.com/item?id=6941135 - Dec 2013 (24 comments)
Fabrice Bellard: Portrait of a super-productive programmer (2011) - https://news.ycombinator.com/item?id=5187585 - Feb 2013 (155 comments)
Fabrice Bellard: Portrait of a superproductive programmer - https://news.ycombinator.com/item?id=2555867 - May 2011 (8 comments)
In the text isn't anything after 2009. And a lot of cool projects happened after 2009.
Ok, 2009 it is!
While the guy is brilliant, I doubt he could fit the role of senior/staff/principal engineer in any one-level-below faang kind of company. Typically, these roles require good communication skills and working together with other engineers (which is really hard). So, while he's very good at the tech level, I think he primarily works alone? In that regard, it would be a very bad fit. I may be wrong, tho.
He is the co-founder and CTO of Amarisoft built on thechnology he developed
https://www.amarisoft.com/
https://www.amarisoft.com/company/about-us
https://bellard.org/lte/
In technically deep domains like Bellard works in, Staff+ roles bias more towards technical expertise, and managers also tend to be more technical and able to more completely address technical coordination tasks. Sometimes we like to assume that if someone is good at one thing, theyâll be bad at something more mundane (to make ourselves feel better), but I sincerely doubt he would have any trouble in such a role.
Staff SWE at a FAANG here.
Fabrice Bellard is not a 10x engineer, he is a 100x engineer. You could attach him to a good people manager and either build a team around him or allow him to work independently on a project that he finds exciting that also aligns with company goals.
I think you are mixing up art, technical skills and productivity.
Put Terry Davis (again him) as senior manager at Apple, and see the result.
From my point of view, Terry has the same level and approaches as Fabrice.
It does not guarantee at all that he is going to be more productive than 100 engineers as you directly claim.
It makes them good in what they like to do (writing obfuscated or low-level code, or implementing from scratch from specifications) as art or creativity.
Thank you for introducing me to Terry Davis. I'm going to read more about him.
I am definitely not talking about art.
When I refer to 100x engineer, I'm referring to the impact that QEMU and FFmpeg have had on the world. I would be surprised if anyone who is familiar with these two projects would disagree that they have been highly impactful.
At M.Bellardâs level one would could hardly even call such an outcome a character flaw, but my occasional privilege of managing - one should rather say, enabling - high performance teams, taught that the Venn intersection of âcompetent with imaginationâ and âcollegiate mannerâ is far from empty, even in the tech sector.
ââWe're delighted to have you here,â he said, âbut a word of advice. Don't try to be clever. We're all clever here. Only try to be kind, a little kind.â Like most university stories, this one is variously attributed and it probably never even happened but, as the Italians say, se non e vero, e ben trovato - even if it isn't true, it's well founded.â ⸺ Stephen Fry.
> While the guy is brilliant, I doubt he could fit the role of senior/staff/principal engineer in any one-level-below faang kind of company.
Maybe but whatâs the point? Hell, I might guess he is terrible at jiggling and basket weaving, too. Complete failure as wrestler, even. But that is kind of neither here or there. Or is it you think staff title at faangs is some kind of pinnacle position every engineer should strive for? It actually always strikes me as a funny title. In college when they didnât have a specific professor to teach or just going to use a grad student they put âstaffâ in the name box so in my mind itâs associated with a random lower rung student who couldnât get away doing just research.
Yeah, staff engineer is a pinnacle "still doing engineering and maybe leadership but not management" position in engineering firms. The academic "staff" is just a "not really one of us" gatekeeping-the-servants title.
> I doubt he could fit the role of senior/staff/principal engineer in any one-level-below faang kind of company.
Why would he want to do that, though?
Yeah and can he do it on a cold rainy night in stoke?
The fact that so many people use FFmpeg and QEMU suggest that he is quite good at documenting, collaborating, and at least making his code remarkably clean and easy to follow. This already puts him way ahead of the average silicon valley senior software engineer that I've worked with. However, he does value independence so I don't think he would have been happy working at a faang-type company for long.
Not really. https://codecs.multimedia.cx/2022/12/ffhistory-fabrice-bella...
>Fabrice won International Obfuscated C Code Contest three times and you need a certain mindset to create code like thatâwhich creeps into your other work. So despite his implementation of FFmpeg was fast-working, it was not very nice to debug or refactor, especially if youâre not Fabrice
Who cares about being a staff at FAANG lmao when he gets to do what he does currently?
Employing Bellard at FAANG would be a tragic waste!
>"In that regard, it would be a very bad fit. "
He might as well be but why would he give a flying fuck about it? He gets to do what he wants and is financially independent for doing just that. Most can only dream about it.
Myself - I do not come within a million miles to his professional level, but I still have managed to do just that - I develop what I want, how I want and get paid for it. I am 64 and still design and develop actively for my own company and for clients. Gives me happiness, motivation to stay alert and more than enough time to still do my hobbies (mostly various outdoor activities).
Is it insecurity about yourself that leads you to baselessly speculate that an accomplished figure is unemployable?
Please don't cross into personal attack, regardless of how wrong another comment is or you feel it is. It only makes things worse.
https://news.ycombinator.com/newsguidelines.html