mpeg

I wouldn't say those two are equivalent. A cryptographic hash requires the exact full document to be available to "recover it" from the hash. With a vector embedding you can extract information related to the document from the embedding alone as long as you know (or can guess) what embedding model was used. You won't be able to reconstruct the document but you will be able to infer some meaning from the vector alone

Where can we find the blog post you made back in January about the RCE fix explaining what measures you took to check if any customer data had been affected?

First thing I looked for... this is an absolutely critical vulnerability that if exploited would have completely ruined their business. No bounty!?

I have a bunch of LinkedIn connections claiming they get those fabled six and seven figure client engagements where the client “doesn’t care about the code, they just care about business results” and a lot of other buzzwords.

I do data consulting part time, I do very much provide business centric solutions and measurable results (alongside code that your tech team won’t hate) and am an expert in marketing technology, applied AI and a bunch of other “hot” subjects like that.

In my work with huge, billion dollar companies, the consulting dream that LI influencers sell feels like a fantasy world. I have to provide and justify a day rate, and when my clients do have a big CAPEX project I’m bidding against Accenture, Palantir, etc. and I don’t expect to win it.

If anyone really consistently lands these amazing, big pocket clients, please let me be your intern.

If you had read the article you would see:

  1/ they need to collect ground truth data for their algorithm to work, it doesn't magically work everywhere. 
  2/ the ground truth data was collected mostly by their clients, it is not their data to give away for free

I honestly don't see a problem with this technology, and I am a huge privacy advocate. First off, it uses the wifi signal strength + a model based on ground truth data to accurately position you in a map. This means that it's entirely opt-in, they can't accurately track you if you aren't using their app / connected to their wifi (yes I know some data does go out to wifi access points even if not connected, but I doubt it would be enough for this kind of tracking, and it can be disabled by the user)

Yes, they mention promotions, but again the promotions would be opt-in – if I use their app to find a product I'm looking for, they might suggest other products along the way that I might also find useful, or they might take me in a route that passes right by them. This is no different to the way retailers stock up their shelves already, placing products next to others you might want, and moving necessity items around when they want to direct you to another part of the store.

I don't know, I think it's a bit harsh to criticise this when the technology has so many applications outside of retail. I would love this in a museum or library, and even in retail I absolutely hate those interactive map displays that modern shopping malls have, where only one person can use them at a time and you have to navigate through 200 store names for the one you actually want to visit

IMHO, the biggest impact LLMs have had in my day to day has not been agentic coding. For example, meeting summarisers are great, it means I sometimes can skip a call or join while doing other things and I still get a list of bullet points afterwards.

I can point at a huge doc for some API and get the important things right away, or ask questions of it. I can get it to review PRs so I can quickly get the gist of the changes before digging into the code myself.

For coding, I don't find agents boost my productivity that much where I was already productive. However, they definitely allow me to do things I was unable to before (or would have taken very long as I wasn't an expert) – for example my type signatures have improved massively, in places where normally I would have been lazy and typed as any I now ask claude to come up with some proper types.

I've had it write code for things that I'm not great at, like geometry, or dataviz. But these are not necessarily increasing my productivity, they reduce my reliance on libraries and such, but they might actually make me less productive.

Repeated strings like that should compress pretty well with gzip, which is one of the core ideas behind atomic css performance, alongside the difficulty to remove dead css – with vanilla html/css your css grows unconstrained

With atomic css, your css stays a constant size, your html will be bigger but is easier to manage as you naturally add/remove html as page content changes, plus compression should be pretty efficient.

I've had to manage legacy frontend codebases with tens of thousands of lines of mostly unused css that were not easy to remove as they might have been used somewhere, that's what led me to start using atomic css.

I don't really love tailwind, but I appreciate it has become ubiquitous so I do use it, and the ideas behind atomic css are solid when you maintain a large frontend codebase. It has also led to the rise of self-contained components as opposed to libraries (shadcn and all the others) which I couldn't be happier about.

Tailwind comes with a set of colors, shadows, typography scale, paddings, etc.

Sure, you can override them, but most people don't and there is definitely a "tailwind look"

This is not correct, the regulation has nothing to do with cookies as the storage method, and everything to do with what kind of data is being collected and used to track people.

Meta is hardly at blame here, it is the site owners that choose to add meta tracking code to their site and therefore have to disclose it and opt-in the user via "cookie banners"

Honest question: what do you do with them? I would be so fascinated to see a video of this kind of workflow… I feel like I use LLMs as much as I can while still being productive (because the code they generate has a lot of slop) and still barely use the agentic CLIs, mostly just tab completion through windsurf, and Claude for specific questions by steering the context manually pasting the relevant stuff

Welcome to tech. Same as it's ever been, I've been a founding engineer twice in my career and walked away with 6 figures in RSUs while the random enterprise sales guy two desks over made 7, and the founder made 9.

I do understand the frustration, especially if you have been affected personally, but it's just the way the world works. Starting a company isn't easy either, and as I'm finding out requires a completely different skillset.

Obviously if there was opportunity for a supply chain attack like that they would have done it in addition to wrecking the IT infrastructure. Regardless, I imagine this will impact day to day drone operation as the Russians might re-flash the firmware to a known good version whether there is a backdoor or not

I use this vscode extension called "python envy" that changes venv automatically depending on which folder of a monorepo you are in, it's great! Wish vscode had it built-in

AI isn't the thing, it's the thing that gets us to the thing

I had a project for a big brand a couple years ago where we experimented with genai and inpainting and it was a huge hassle to get it working right, required a big comfy pipeline with masking, then inpanting, then doing relighting to make it all look natural, etc.

It's crazy how fast genai moves, now you can do all that with just flux and the end result looks extremely high quality

It’s not just verifying that the address exists, for KYC you usually check if the address is linked to that person, and to do so you need some way to account for variations in the way the address is written.

I have done it as a hiring manager, it's really not that hard.

1. You can use an employer of record service which costs a few hundred bucks a month – it seems like a lot... but if I'm already paying a recruiter £12 to £25k to find me a senior data engineer in London on £80 to 120k that is going to want to WFH 3/4 days a week, I will gladly pay £400/mo for an EOR service

2. You can also not hire them, and use their services as independent contractors instead. I've never had an issue doing this with my finance teams, as long as the contractor submits a valid invoice they don't care who they are. Plus, it's good for cashflow (net 30 to net 90 is pretty standard) and the hire gets a nice tax save on their end.

I do understand that at large companies it can be tricky, but IMHO at startups there is little excuse. I suppose it all doesn't matter if you're playing with unlimited silicon valley VC money, I've only ever had to deal with european investors and they love a bit of smart frugality.

I don't doubt he's in the 1% or 0.1% of candidates you're interviewing, but there is one very simple solution startups could apply to make it easier to find top talent -> remove "US ONLY" from their job listings.

That's a particularly terrible cold email, you can tell he didn't even bother applying some basic personalisation to it outside of [COMPANY_NAME]

The cultural/political issues are deeper than you'd think, there are some within Catalonia that would claim all of us Catalan speakers as part of the same cultural umbrella, negating our own cultural background as derivative.

That's why I say that I personally (as a Mallorcan) would much prefer the language to be associated with a Spanish flag than a Catalan flag. Some people would agree and some would disagree – ultimately the best thing is to not link flags to languages anyway.

To give you another example from Issen, they're also using the American flag for English which is probably even more controversial :)

Catalan, the language, does not have a flag. As a native speaker (of a dialect) who is not from Catalonia I would much rather the language be represented by the Spanish national flag. Although people from Andorra might also not love that.

Languages and flags don't mix well.

You'd think that, but a globally installed npm package is annoying to update, as you have to do it manually and I very rarely need to update other npm global packages so at least personally I always forget to do it.

A bit on the nose that they used a sample from a professional voice actor (Jennifer English) as the default reference audio file in that huggingface tool.

Students are the only people with the patience for deep RE, I spent hours and hours in my teens unpacking binaries that used similar VMs and got pretty decent at it.

Nowadays, there is no way I could do it, I tried to get back into hackthebox recently and the new RE challenges make my brain hurt.

It's a very common term in reverse engineering, and low level programming in general. In VMs you usually say "stolen" to refer to bytes/instructions/constants that have been taken from the original binary and put somewhere else (whether obfuscated or not, whether still in the local binary or in a server like with denuvo)

That's both genius and a massive hack, love it.

For those who are wondering: they alias @typing.overload as @ovld only for type checkers.

Normally, @typing.overload is a way to define multiple function signatures with empty bodies so your type checker can work nicely with functions that for example only return a certain type when called with a certain signature, but the function implementation is left to you and usually involves a bunch of runtime type checks to branch your logic the right way

Yeah I doubt there's any AI here, they might even use other tools like say YC's up.codes for the compliance with local codes, then they arbitrage cheap labor at a premium price because "AI"

And the testimonial photos, they're all AI generated

So, I know people that have done exactly this in the UK, but they did it anonymously. Might be unethical but depending on your employer you can YOLO it. Basically you have a close friend set up a LTD which takes a day, they are listed as sole director. You then draft up a contract agreement with your friends company.

After that you’re ready to champion the sale internally, you pitch it as something a friend has set up that might help with X project. Your friend is the main point of contact and gets to keep a % margin through the LTD

Of course I would not bother with this unless it’s 6 figures a year, I know a guy that bootstrapped his startup this way when he was working for a large co

The real killer feature of lottie is that there is an after effects plugin that exports as a lottie JSON. It doesn't always work 1:1 especially with things like AE filters on certain layers but when it works it saves a ton of time