#eMudhra

kinda only putting feelers out but, anyone looking for audio ML experts, or anyone looking for ML experts who works in the BCI area with some, though no recent, BCI experience? EU based, currently in Tallinn, willing to relocate, *especially* willing to relocate towards germany from here.

I have shipped SOTA ML audio models to 100M+ edge devices, owning everything from data collection to training, integration, rollout and live metrics / monitoring, working with other internal teams where needed. Most recently, I've created and shipped a model that improved the core Customer Dissatisfaction metric in our strategically most important product by 5% relative (so far - rollout ongoing). I usually work in Python mostly, C and C++ sometimes, I have hobbyist level Rust experience. I have extensive experience with using LLMs for coding work and understand what they can and cannot do pretty thoroughly and where the real bottlenecks are, and can work with or without them.

:boost_ok:

Bit annoying that gifs don’t quite work right in Mastodon embeds

I don’t think i‘ve ever seen someone quit a job this hard

waoh fast man in a fast car

it‘s the building I watched getting blown up 3 days ago!

Okay, time to check it out, then

you know that one clip? "I am not a number!" etc?

I wonder if the thing it is from (TV Series, "The Prisoner", 1960s, from the UK) is actually good

apparently it's on YT for free https://www.youtube.com/watch?v=ajvdrWJSqdU

I think one thing bluetooth desperately needs is a programmatic way to tell the device: "okay, I know that whoever built this is very proud of all their ADVANCED AI AUDIO PROCESSING but listen to me: turn all of that off, and give me the raw audio signal"

like for BT audio the list of improvements that would be genuinely amazing are:
1) fullduplex fullband audio by default and banish the "communications profile" to the shadow realm
2) let programs turn off the maximum viable amount of "on device" processing

and everything else is basically whatever. "spatial audio" whoooo gives a shiiiit we're failing on devices consistently working reliably when you make a call

are they finally making bluetooth audio good (probably not but hope dies last)

apparently we‘re on woke 3.0 now, so I assume that soon we will have woke for workgroups 3.11, and later woke NT will finally bring proper preemptive multitasking support

So I take it the Americans didn’t cock it up while I was asleep

the countdown, has begun

https://halcy.de/padoru/ HASHIRE SORI YO

Apparently a draft for letting people control animated image play state exists, at least. So if that passes then in 5 to 10 years after that it might become possible to actually do this in a normal way

no, of course not

https://jakearchibald.com/2024/video-with-transparency mmmmaybe this works in 2026? maybe apple have gotten their shit together? lets try

and in terms of video formats because I guess that's what you're supposed to use instead, of course apple can't be bothered to be normal, as usual

sssssigh

how is it that in the last god knows how many years google has deigned to add like 15 billion pieces of user hostile crap to HTML but there still isn't a way for me to control animated image playback

anyways I certainly hope that the people who do infra for the OTW and AO3 know how much they are appreciated and how much they are doing something that no one else can and how much it matters to a lot of people. absolute best of luck.

honestly, sad times. in any other year, this would be a clear favourite for the Funniest Security Incident Of The Year Award. But this year, 4 is sweeping it, there's just no question. the GTA VI of Security Incidents.