Cixelyn

Spellbrush | Tokyo, JP or San Francisco, CA | Full-Time | Onsite

We build AI models to make anime real.

- Game Dev: looking for engineers who know unity, love the anime aesthetic, and want to work on our upcoming strategy RPG.

- Full Stack Software Engineer: looking for smart generalists who love anime and want to work on advanced data processing pipelines for petabytes of anime images & video.

- Audio/LLM Researcher: looking for senior researchers interested in working on foundational end-to-end audio models. Must tolerate^Wlove high-pitched moe anime voices.

Ping over name of best waifu or husband to jobs@spellbrush.com. Can sponsor visas to either USA or JP.

See all roles over at https://spellbrush.com/careers

Thanks for the shoutout! Boring careers pages have long been a pet peeve of mine.

We also work on anime image diffusion models, and it would be massive shame if our own careers page didn't highlight the results, haha.

Spellbrush | Tokyo, JP / San Francisco, CA | Full-Time | Onsite

We build AI models to make anime real.

* Game Dev: looking for engineers who know unity, love anime, and want to work on our upcoming strategy RPG.

* Full Stack Software Engineer: looking for smart generalists who love anime and want to work on advanced data processing pipelines for petabytes of anime images & video.

Ping over name of best waifu or husband to jobs@spellbrush.com

See all roles over at https://spellbrush.com/careers

Spellbrush | Anime AI researcher, Software Engineer, or Unity Engineer | San Francisco OR Tokyo | Sponsors Visas

Spellbrush (YCW18) builds state-of-the-art AI foundational models for anime and video games.

We probably spends 10x more compute on anime than anyone else in the industry - we think our results over at nijijourney.com speak for themselves.

Looking to both expand our game development team and our AI research and data processing teams.

JAX and TPU experiences are a plus.

Otaku looking for a job in AI are welcome to ping over name of best waifu or husband to jobs@spellbrush.com

you check in the uncompressed audio, and then fmod, wwise or whatever other middleware you're using will compress to a different format and compression ratios on a per-platform basis during the build process (i.e. vorbis on PC, AT9 on PS4, FADPCM on mobile, etc.)

Spellbrush | San Francisco or Tokyo | Onsite | Full-time

Looking for anime fanatics (read: weebs) who want to work at the intersection of AI, anime, and games.

Looking for React experts, Unity experts, GPU Infrastructure experts, and Pytorch + JAX experts.

Especially interested in folks who want to push the boundaries on training large-scale foundational diffusion models.

Hiring for all roles. If you're sure you can help, but don't see a position that fits, DM me via LI or twitter (in bio) Include name of best waifu/husbando to bypass spam filters.

https://spellbrush.com/careers

The Live reconciler[1] included in this project looks quite interesting as a reconciler for building a canvas-based image editing app.

Since WebGPU is still quite experimental and not usable without install canary browser builds, I'm really hoping that Live can be used separately. Like the author, I'm pretty sick of doing the data retained/declarative dance with libraries like pixijs many times now.

At least from this article, it seems that the author is using their @use-gpu/state[2] package independently. Would love to know if anyone might have any more details or experience?

[1] https://usegpu.live/docs/guides-live-vs-react

[2] https://usegpu.live/docs/reference-live-@use-gpu-state

One of the Spellbrush team members here! It's always great to see niji・journey[1] being used in such creative fashions!

Happy to answer questions or otherwise that folks might have.

Also props to coveragecat. The generated cats are _super_ cute. Especially the ones that made it to the current landing page.

[1] discord.gg/nijijourney

Network Information Center. They did DNS stuff before ICANN

See: https://en.wikipedia.org/wiki/InterNIC

is anyone running a large baremetal distributed minio cluster in production? would love to know how reliable it is

would love to see the riser setup that you're using for such a monster!

we mostly gave up and just got barebones machines since the cabling situation becomes pretty tricky, and the barebones total cost is low relative to the GPUs anyways.

dev here: the characters aren't tradeable atm.

but something we do allow is for players to create characters on the main waifulabs site[1], and then share QR import codes of that custom characters with friends.

[1] https://waifulabs.com

No team page yet! Friends with near, but haven't had a chance to meet gwern yet (maybe one day??)

Arrowmancer is really our first attempt at scaling it up; hoping to do even cooler generative AI-related in that production.

Will probably blog probably in a week or two.

But the tldr; is a lot of scavenged EPYCs and nvidia GPUs all in a large sound-proof rack.

San Francisco! Just sent over a ping!

Right now it's a small team of 6 people, and we have a bit of funding + compute credits to train models. There's a bit of revenue from some past projects and AI-consulting, but we're mostly betting big on our new AI-powerd mobile title Arrowmancer[1].

I think the effort required for such an undertaking is vastly underestimated by readers.

Haha for sure. Hosting a real-time ML model for people to do sub 1-second inferences at HN-load scale is definitely nontrivial.

[1] https://arrowmancer.com

Here's our current game trailer! https://www.youtube.com/watch?v=8WvRgb6kh4s

(link for the lazy: https://youtu.be/Pab8pG5WbXQ)

Thanks so much! It's done by our fantastic animator[1]!

GANs are quite interesting and we didn't see many approachable explainer videos targeted at lay people, so we decided to make one ourselves!

[1] https://twitter.com/bumblingbeebo

Anime Mobile Games | Frontend, Fullstack, Gamedev | San Francisco, CA | Fulltime | Remote OK

We're a YC startup working on an ~Anime Mobile Game~ powered by state-of-the-art deep-learning illustration AIs[1]!

Give us a shout if you like anime and are an expert in any one of our three open positions: Elixir, React Native, or Unity!

Official job listing is here[2] but also feel free to email jobs@spellbrush.com directly or twitter DM @cixelyn.

Make sure to put name of best waifu/husbando in the subject line so we know it's not spam :)

[1] https://www.arrowmancer.com

[2] https://www.workatastartup.com/companies/spellbrush

Yeah, this is a pretty narrow window of the millennial generation. If you had been born too late (e.g. Gen-Z), you'd have high production AAA-production mobile games to play between classes and after school instead.

One thing the author overlooks is how game programmers were able to exploit the primitive networking capabilities of the calculator. I fondly remember playing multiplayer Bomberman by plugging two TI-83 together via the 2.5" link cable.

I recently had enough nostalgia to pick up a new 2021 TI-84 Plus CE PYTHON unit. On one hand, it hurts to buy a 20+ year old CPU for over $100. On the other, there's something satisfying about still being able to program TI-Basic via muscle-memory using the token-based input.

Curious -- any reason you didn't just go with a single machine export + expansion disk shelves on something like ZFS? Installing a MinIO gateway would also act as a bare drop-in for S3 too.

Asking since we're in the same position as yourself w/ high double-digit disks trying to figure out our plan moving forward. Right now we're just using a very large beefy node w/ shelves. ZFS (via TrueNAS) does give us pretty good guarantees on failed disks + automated notifications when stuff goes wrong.

Obviously a single system won't scale past a few hundred disks so we are looking at alternatives including Ceph, GlusterFS, and BeeGFS. From the outside looking in, Ceph seems like it might be more complexity than it's worth until you hit the 10s of PB range with completely standardized hardware?

Spellbrush | Frontend Engineer | San Francisco, CA | Full time | Remote or Onsite

We're a small, tight-knit team of MIT-trained researchers applying the latest advancements in deep learning for *anime* and *video games*.

Currently hiring a full-time front-end engineer to help us launch the world's first AI-illustrated game.

You should definitely reach out if you:

  a) like anime
  b) have experience with react and/or react native.

Make sure to put name of best waifu/husbando in the subject line so we know it's not spam :)

[1] https://www.workatastartup.com/jobs/43940

Yeah! Docs start on page 323 here: https://datasheets.raspberrypi.org/rp2040/rp2040_datasheet.p....

Looks like there's an assembler for programming the PIO peripheral; they seem quite capable with two IO registers and two scratch registers.

I've used a similar sort of embedded state machine inside EFM8 microcontrollers before. They're great not just for implementing custom communication protocols, but also for offloading simple tasks: one example I've coded before is that you can offload software key debouncing to the state machine, and even get an interrupt on keypress. This greatly simplifies the main loop as you don't need to dedicate polling time or write debounce calculations!

EdgeFS is one I was looking at recently

Do you have any additional info? EdgeFS's github[1] doesn't work; does repo access require a Nexenta sales call?

We're also looking into asynchronously replicated FSs, I think built-in caching + tiering is slightly nicer than cron + rsync; would love to know what other solutions you looked into.

[1] https://github.com/Nexenta/edgefs

Now that nvidia owns a full switch OS, I wonder if this means that Onix, which they inherited from their Mellanox acquisition, will eventually be phased out.

We’ve been running Onix just because it’s cheaper than Cumulus (read: no yearly contract).

According to our creative director, the Surface’s digitizer is significantly worse than the Wacom one. iirc the first generation Surface licensed Wacom technologies, but in subsequent generations Microsoft switched to their in-house nTrig which they got through an acquisition.

In fact the Wacom is better by a large large enough margin that she’s willing to forgo using the (reasonably well-designed) Surface for the bastard child of a laptop that is the Wacom Cintiq Mobile Studio Pro 13.

I think the paper that the NYT article is referencing is actually this one, published yesterday: A Framework for Intelligence and Cortical Function Based on Grid Cells in the Neocortex

https://www.biorxiv.org/content/early/2018/10/13/442418

Jeff Hawkins is scheduled to give a talk at the Human Brain Project Open Day[1] in Maastricht, NL about this paper.

The one linked in the parent is from last year, related to the mentioned story of him running his finger across the coffee cup.

[1] https://www.hbpopendaysummit-2018.org/programme/keynote-spea...

Build.moe | Designer / Illustrator, AI Researcher | San Francisco / Half Moon Bay | Full-Time | Onsite

I'm the founder of a small YC-backed startup building a cutting-edge artificial intelligence that helps artists draw and paint faster. Our primary customers are in the animation and games market, but we’re finding new applications in the space every day.

--DESIGNER/ILLUSTRATOR--

We're looking to hire a full-time designer/illustrator with a passion for anime and board games. Ideally, you've done visual design for games before, you know how to design for print. INTERNS for this position are also welcome.

Please check out more information about our design position here: https://build.moe/hn-designer/

--AI RESEARCHER--

We're also looking for AI researcher to help with the development of state-of-the-art generative networks for artistic work. If you want to work alongside other world-class researchers, and you have some experience with GANs, please drop an email to jobs@build.moe with the name of best waifu/husbando as the subject.

The Nvidia one looks quite a bit nicer than your average eBay ruler though... full 12", two color silkscreen, ENIG finish, and some crazy tiny-pitch BGA footprints. The component layout is also not haphazard like most of the other rulers I've seen; it's much more of a work of art.

I definitely want one.

Other JVM languages are technically allowed, but AFAIK they generate way too much scaffolding & reflection-based flow such that you end up burning your entire bytecode computation budget just in calling one or two functions.

So yes, you could probably use Jython or Jruby, but you wouldn't be competitive against the people writing straight Java.