We were not accepted into Google Summer of Code. So, we started our own
qdrant.techIn general it seems that organizations that are accepted to the GSoC program are no-profit like associations, that develop a product for the sake of developing it, probably there are also exceptions.
Anyway, it seems that qdrant is more a for-profit organization. So maybe that was the one of the criteria that was taken into account to exclude it?
My understanding might be wrong/incomplete, please let me know if that's not the case.
This is not a reason. We participated successfully last year and even hired one of the guys afterward.
They tend to rotate organisations. It’s not necessarily you.
Yes. This makes sense. There is no criticism of Google at all. We just want to continue supporting contributors.
No mention of how much you will be paying contributors for the different tasks in your blog post.
You should include a note in the blog post or at least add it to Notion, beneath the time estimates.
.
Did you look a bit deeper? Qdrant is written in Rust from scratch. The project is over three years old and started long before the "gold rush." :) Run it wherever you want, in your data center, locally, with no limitations. https://github.com/qdrant/qdrant-helm
This is what Haskell did last summer: https://summer.haskell.org/news/2024-01-20-summer-of-haskell.... Fortunately they got accepted this summer.
strange that such an old company would participate
FWIW most organizations aren't accepted every year -- they have a limited number of slots and they like to use some of those to invite new organizations in each year.
So don't start looking for reasons why any organization wasn't accepted; there probably isn't one.
So they didn't find potential hires that would work for free on improving qdrant without any certainty of employment, and now they are starting a competition for it (for free again)?
From the article: "we are introducing our own program with a stipend for contributors!"
I also applied with Glicol (https://glicol.org/) and got rejected. I guess the main reason is that the project is not as mature as others. I am basically working this project on my own with almost zero extra funding. There are so many places I want to change.
I am currently working on a new website. The old stack is Vite, Svelte and Windi CSS (discontinued unfortunately). So this time maybe Astro + Solid + Tailwind.
And I am also trying to rewrite the whole Rust backend if possible, so there is quite some work to be done. What I want to change most is to make the dsp algorithm of each node clear and easy to understand and contribute to. And I also hope that the entire rust project can have complete bench and test, as well as ci, and get rid of the proc macro.
Generally speaking, what I actually care about is how to compose music, and the new possibilities that live coding brings to improvisation and composition. There is also network cooperation, real-time or non-real-time cooperation, and cooperation with AI. What possibilities can these bring?
Let me know on GH or Discord if you are interested.
It's a good chance to try Rust, WASM, DSP, etc.
Glicol is a great project. Good luck!
> I am currently working on a new website. The old stack is Vite, Svelte and Windi CSS (discontinued unfortunately). So this time maybe Astro + Solid + Tailwind.
It's a bit amusing to read that Vite+Svelte can be considered an "old" stack. :)
I'm curious: what makes you want to move to Astro+Solid?
> WASM-based dimension reduction viz
> Implement a dimension reduction algorithm in Rust and compile to WASM and integrate the WASM code with Qdrant Web UI.
Easy, just use a Rust crate to fit a PCA (https://crates.io/crates/pca), then at runtime do a matmul between the fitted matrix and the embeddings to get it reduced. :P
Speaking of which, there's a surprising spike in downloads for that crate on the date this blog post was made.
It's not as simple in practice, and even popular dimensionality reduction techniques like UMAP require you to reference the original dataset which is infeasible for large datasets. The hacky approach that would be good for production use (maybe not "just want to visualize 2D embeddings because they look cool") would be to train a small Parametric UMAP model (with likely a non-Rust implementation: https://umap-learn.readthedocs.io/en/latest/parametric_umap....), then convert the trained model to ONNX for WASM.
“Easy, just…” lol I do like the joke (and I appreciate you pointing out that in reality things aren’t necessarily _that_ simple), but as I understand it these programs (internships?) are targeted mostly at young folk/undergrads who are still learning the core skills, in which case projects that are conceptually easy but require you learn the skills to actually implement the conceptually easy task within the confines of real world constraints feels reasonably appropriate?
I could be wrong, because it’s not really a field I do anything in but based on what you said it’s sounds about right for a first “real project” task an intern would do at a big tech summer internship (this is as someone who has helped summer interns over the years, but never did one myself so it’s based on seeing what tasks are given to someone in their first ever internship vs people who have interned previously)
Just dropping my $0.02 cents on the Jepsen testing for Qdrant's distributed guarantees. While Jepsen is a solid choice, have you considered throwing Antithesis [1] into the mix as well? If it's already on your radar, then no worries, just figured I'd mention it.
What's with the hashtags inside a normal blog post? It's not like they're writing this on Twitter or anything, and the hashtags aren't clickable, so I wouldn't expect this to serve as SEO. I would guess one of two things: (1) they plan to post the same content to social media or (2) they've decided that people search hashtags in their search engines in general.
I'd be interested if anybody has more insights into this.
Appears the blogpost text is from a LinkedIn post (where the hashtags are clickable): https://www.linkedin.com/posts/zayarni_qdrant-summer-of-code...
Nothing more interesting than copy-paste I'm afraid
Same reason they have random emoji at the ends of each bullet point: it's an intentional forced-quirky vibe.
It has been a part of most major AI open source projects lately, for some reason: https://sigmoid.social/@minimaxir/110951886465291229
I agree that people use emojis waaaay too much for this sort of stuff (see also: certain projects' commit messages, various READMEs).
Imagine getting upset over emojis.
It’s just injecting a little bit of life and visual variance into the text.
>It’s just injecting a little bit of life and visual variance into the text.
The best way to inject a little bit of life into a text is with engaging and vibrant writing, not putting colourful stickers all over it.
I guess the sticker thing works for 6 year olds though.
Good writing is nice, but emojis can certainly make it colorful.
Personally I feel the (excessive) colors can be distracting, but that's just my view.
Sorry, my bad. Was partially copied from Linkedin, where I post most of the content. No idea behind. The emojis are from here https://qdrant.to/gsoc24
Or (3) hashtags have become some intrinsic quality of writing by people whose minds have been entirely corrupted by social media junk
This is the first GDPR banner I've ever seen that gives no options but "Consent" or "block our cookies with your browser." Is this willful noncompliance?
The GDPR does not require a pop up or banner… but it does require explicit consent through opt-in (or no tracking). So this is literally worse than nothing.
Lodge a complaint with your national DPA.
Who would spend even a minute for a unknown company called 'Qdrant' ?
They would develop for Google because Google would give an additional value to their CV.
This is an incredibly dumb comment that you should delete. Anyone even remotely aware of RAG, LLMs and vector databases has heard of Qdrant.
That was a dumb comment, but you are wrong. I am remotely aware of LLMs (how could one not be, in this day and age) and I have never heard of Qdrant.
I knew putting LLM in that list of pre-reqs would result in a comment like this. Oh well.
I am remotely aware of all the three things you listed (and have experimented a bit with them) and I haven't heard of them.
Once a comment in Hacker News gets replies, it cannot be deleted anymore.
Yeah I think you’d take the GSoC so you could have qdrant and gsoc on your cv, but also many folk do need a job to pay bills, and even an “unknown” company paying bills and being experience on your CV is better than Walmart (in my case The Warehouse)
[edit: s/Qurans/qdrant, sigh autocarrot]
How much are they paying? It is $current_year and at some point we should demand that all job postings (this is a job posting) should come with salary ranges at a minimum.
Are they even compliant with the law with this post?
https://developers.google.com/open-source/gsoc/help/student-...
It’s not a job. It’s an incentivised open source contribution program for college students.
When other organizations have done this they just pay the same as Google.
> we should demand that all job postings (this is a job posting) should come with salary ranges at a minimum
Agreed. Not even considering jobs that pay under 120k, not worth my time. The literal only reason to hide the range is so that you can lowball people
I have no idea :-/
They recently raised a Series A: https://news.ycombinator.com/item?id=39101682
Additionally, the product that this Summer of Code is for is open-source so it's a win-win for everyone.
I wonder if Google wants rights to what is developed in GSoC.
If so, that might explain the lack of invitation.
> I wonder if Google wants rights to what is developed in GSoC.
They don’t. I participated in GSoC working on something that competes directly with Google (LibreOffice) and was never asked to assign copyright or anything like that.
IMO GSoC is a relatively cheap way for Google to get some goodwill and boost their brand among college students; it’s not really a core part of their competitive business strategy.
That's very cool then. Thanks for responding.
> ... unknown ...
This is a resounding instance of "tell me you don't know the domain without telling me you don't know the domain" and I think you'll find them interesting if you look into it.