I'm looking for inspiration, so what's something really exciting that you're working on? (day job, hobby project, whatever...)
It's just radically freeing to not have to wait for a boot, to have direct control over hardware resources like the frame buffer and audio registers, but to start with something as simple as Hello World and go in either direction from there — up in programming complexity or down in systems understanding. I'm very close to getting it into friends' (and their kids') hands. I'd love for it to take off, and have experience actually selling services and products, so I don't think it'll just be a hacker toy, but who knows? Regardless, it's been probably the single best learning project I've ever done. And it's so fun! It'll continue to be so for me whether it sells and/or has a community that builds up around it, or if it's just my passion.
"Radically freeing" is exactly how I'd imagine it too.
If you need any help at all on the non-technical aspects (logistics, fundraising, manufacturing/stock-keeping), please shoot us a message (email in bio). Not asking for anything in return, would just love to see something like this come to market.
So many creators overlook these aspects, and they're arguably more important than the technical specifics.
I'm sure it would reach the front page very quickly, and I'd love to buy one if only to support you!
Creative writing releases psychic energy that activates machines / opens doors and all sorts of other things. It also exports .txt files.
Depending on what you write, you'll learn different sigils you can arrange into a circle that will activate when you're in danger. Writing a romance novel gets you different powers than a horror or sci-fi short story, for instance.
https://twitter.com/LeapJosh/status/1459527876118814728
It's pretty weird, but I'm really proud of all that I've done myself this last year. Solo indie gamedev is a perilous, extremely stressful, maybe stupid career choice but I'm realizing a lifelong dream.
I also think I might be inventing/innovating a new genre of game so it's fun to maybe leave a mark that way too. I hope it'll squeeze some creative writing out of people they never would have written otherwise.
My clients are mostly local (i.e. Switzerland) acquired through word-of-mouth. Client acquisition is a challenge because I have to juggle sales and execution. However I am super excited to be in this field because I believe this is exactly what precision medicine is.
The company is called YugaCell (http://www.yugacell.com)
https://meomix.github.io/antfarm/
I found some code written in 1991 for Unix (in C) and decided I would port it to React and get it running on the web. I have to implement persistence and iron out a few bugs, but it's at least functional at this point.
I wrote the original coder and let them know I'm rewriting their code 30 years later. They're stoked to see it come back to life and I find that motivates me to do a good job. Plus, some of my friends' kids have started to take an interest in it.
I'm not sure which direction I'll go with it long-term. I kind of wanted to get it running on an e-ink screen and have it sit on my desk or become more like a tamagotchi. I don't have any experience working with hardware and think it would a cool way to learn some of that stuff.
I'm probably going to add some more features to it like a queen ant, food, pathfinding, etc. I saw this cool "ant-based solution to the TSP" https://www.theprojectspot.com/tutorial-post/ant-colony-opti... which I could see being solved by the ants over time and then introducing variables like weather or user-interaction which disrupt the path and then watching it get solved again, etc.
idk! I just played SimAnt when I was young and impressionable and am taking it way too far now. :)
It's interesting in that it's very different from tech, but also rewarding to make a difference in people's lives.
It's often possible to write in, rather than attending hearings in person. Get a mailing list of people and have them write in.
I have a really good mental model of superposition. I'm pretty sure I understand quantum annealing. I'm working on the bloch sphere and quantum gates.
Once I'm sure I understand things, I'll write up some code to simulate it, and then see if that code matches up with the stuff already out there, result wise.
I'm consciously guarding my ignorance of how it is simulated, lest I fall into the same mode of thinking about it as everyone else.
I think I can simulate quantum annealing with an almost trivial amount of code.
- in-browser audio suite - https://severak.github.io/cyber-music-studio/
- my very own social network - https://kyselo.eu/
These are my major software hobby projects. Social network one is more useful than interesting. Audio suite is very interesting but not that useful. I hope that openstreetmap renderer can be the middle ground - both useful and interesting.
So I built BuzzMeIn that hooks up to old school phone based apartment buzzers via a Twilio phone number and now the sky's the limit of what I can do.
Plus it solves real people everyday problems, making life more convenient (which I'm also a big fan of)
Then I’m building that into a platform where people can fork any query, modify, and publish with their own analysis in order to build a portfolio.
My first market is sports data. There are many aspiring analysts, and I want to 10x the number of people who do this work. And I think the best way to learn Analysis is SQL, and the best way to learn SQL is by building off other peoples queries (learn by example / exploration).
This is what OLAP-like engines are built for.
When you have these types of queries, the relational model ends up degenerating in a star schema with queries issuing a join for each data column on the first projection, and then a pass on that projection for aggregation, typically working on a time range that's relatively recent.
For these, native columnar stores are usually a better option. Things like Apache Pinot https://pinot.apache.org/ might be a better fit.
If you add a real-time requirement, it gets even more challenging, going into the realm of custom built query engines, such as those that back products as those built by Medallia or other customer experience companies.
It's a really interesting niche.
Since I have you here, what do you think about MADlib from Apache? Basic ML in SQL is really appealing - one of my objectives is to teach people analysis using SQL, and it'd be great to only need to develop that one skill (rather than "learn SQL and python"). https://madlib.apache.org/
I envision a future 20-30 years from now as profits dry up the industry collapsing and leaving municipalities holding the toxic asset of an expensive to cleanup site. These sites will then require municipalities to write grants, hire construction staff, etc.
This is where I'm at now: https://search.marginalia.nu/explore/random
I'm leaning toward some sort of categorization or tagging system, but trying to figure out ways of designing it that are inherently resilient to sabotage and manipulation.
[0] https://quickz.org [1] https://news.ycombinator.com/item?id=29836852
I'm using plv8 to set up a deno-type runtime in postgres lol. It's silly but so fun and weird-feeling to me.
It's very finicky right now and most build tools are not ever going to work like this obviously but I have a little tiny "server" running out of postgres just using js tag functions for templating.
The form factor was intended to allow for portability (it has a handle and I can carry it around with ease, and the Pi can run off of the battery when then AC supply is disconnected) and to have enough space to allow for positioning multiple cameras for binocular vision and multiple microphones for stereo "hearing", as well as room for the other random sensors. Oh, and I'm re-using the existing speakers for the audio out so I can work on integrating speech synthesis into the whole kit and kaboodle.
For the software, I'm looking at starting with a super-simple BDI interpreter for the basic "cognitive loop", using an RDF triplestore for semantic knowledge, neural network models where appropriate (object recognition for example), and then start trying to build up from there. Also looking at systems like SOAR, ACT-R, OpenCog, etc. for inspiration.
Of course most of the "heavy lifting" from a computational viewpoint will need to happen on a server somewhere, since a single Raspberry Pi can only do so much. So there's a corresponding server backend piece that will work in conjunction with the remote portion. For now it'll be a fairly low-end server that's physically here at home, but if/when I start needing to scale things up I'll probably switch to using cloud resources on AWS as needed.
All in all, the basic idea is to have an "AI bot" that is alert, observing, and (hopefully) learning all the time and that will learn more like a child learns, compared to the way we train ML models today. That's not to say that there might not be some batch mode training as part of this but I'm hoping to experiment mainly with learning modalities than can happen in real-time. There's no fully fleshed out theory that I'm working off of, but I plan to tinker with a variety of things - Hebbian Learning, Reinforcement Learning, etc. Maybe I'll learn something interesting, maybe not. But it should be fun in either case.
It is called https://shepherd.com/
I launched about 10 months ago we are slowly getting there :)
How does it work?
I ask authors to recommend five around a topic/theme (so every book on the website is personally recommended by an author who is passionate about it). So you get some great recommendations on things like:
The best books on artificial intelligence that are not full of hype and nonsense https://shepherd.com/best-books/no-hype-and-no-nonsense-arti...
The best books that tell a cautionary tale about world-changing technology https://shepherd.com/best-books/cautionary-tales-about-world...
The best books whose dystopian visions were eerily prescient https://shepherd.com/best-books/dystopian-visions
Then, I build out bookshelves (aka topic pages) using NLP. This is very new and only 30 days old so I am still improving the engine / topics. But, it is all tied to Wikidata so you can search via Wikipedia topic (and some other cool stuff down the road).
Bookshelf on artificial intelligence https://shepherd.com/bookshelf/artificial-intelligence
Bookshelf on neuroscience https://shepherd.com/bookshelf/neuroscience
Right now I am working to roll out a big improvement to the recommendation engine. And, then to integrate book genre data, which is a massive project. I want to be able to go to the World War 2 section and say "show me all historical fiction", or on the AI bookshelf to show me all "science fiction".
Let me know what you think :)
Think rss, but for my personal feeds.
Still mvp, lotsa bugs, but if anyone has similar problems -> https://fetcher.page
It's pretty wild that it's possible to make an open source 3D MMORPG where everything is procedurally generated and combat is in real time. That's a combination of words that shouldn't work.
It sounds so horrible, I can't get my mind off it! :P
The idea is to tag files/documents with attributes like "auto", "insurance", "2022", encrypt the file locally (using your manually entered key), copy it to a network share/cloud storage/etc., and add the tags and document information to a database.
Later on, you can search for documents based on the tags and get a decrypted copy of the file.
A DB to contain all the metadata, and blobs or object storage for the content itself (perhaps with backup metadata). The database manages the namespace, object storage manages the storage layer.
You can easily and quickly search file metadata etc, there could be multiple copies of the content, perhaps distributed, etc, etc.
remote storage with local indexes, where the remote storage only holds encrypted data (can be postgres or s3 or whatever), and the only way to make sense of the data is if you have access to the local encryption keys and indexes.
The local indexes would function like your notion of tags, and also be somewhat encrypted (hashed probably) so you can find your data locally but only pull it from the remote when needed.
https://polygonjs.com/particles-music
And it's done with a visual node-based editor I'm working g on. Here is a tutorial of it in action.
https://m.youtube.com/playlist?list=PL2WfjN_pVGgj32-ZJc_VNaZ...
I'm currently exploring how AI/Natural Language Processing can automatically organize personal knowledge and notes by tagging and linking ideas / content together [0], removing the friction of organization to help you find new connections and ideas.
If you're interested in this domain, I'd love to talk to you!
I made a discord server in response (we mostly idle) to discuss knowledge management tooling
Awesome Knowledge Management https://discord.gg/XPNeDSQE2j
Would love to continue to chat about existing tools and new ideas in this space!
The goal is to provide a library that’s autogenerated from your DB schema that can express any EdgeQL query and statically infer the result type.
I’ve done a lot of digging into various SQL query builders for TypeScript, and while some noble efforts have been made, there are structural issues with SQL-notably the fact that it’s tricky to write queries that return nicely structured results-that kneecap those efforts.
By starting with EdgeQL as our target query language, The problem gets more tractable. It still required some truly nextlevel TS wizardry. I’m the developer of Zod and tRPC too, so it’s not my first rodeo, but this is the thing I’m most proud of. Not to sound too grandiose, but this really does represent a “3rd Way“ beyond the usual “raw SQL vs ORM” debate.
Also got into 3D printing, using OpenScad for creating the models. It is almost magical when you can think of something that you want, but they don't make, yet you can make it yourself (different brackets for holding accessories on a bicycle, clips that holds a plexiglass screen in front of the TV to protect it, etc, things like that).
These are a couple of ideas of things that a technical mind can work on (woodworking and 3d printing both require some amount of precision and design). Keeps things interesting.
Users create memes and enter them into their preferred tournament(s). Other users on the app are invited to judge the memes. Eventually a winner is chosen, and they can win a real prize like a digital gift card. Tournaments are free to enter for all users. Each user is given free daily tokens to enter as many tournaments as they want.
Users can also host their own tournaments (for a fee). Users can be regular people, social media influencers, and companies big and small. Users can give away whatever prize they want.
Web: https://planetmeme.com Social links on the website.
It’s been fun to work on, plus it’s a challenging technical problem: in order to scrape job posts for many companies, you need to make a very generic scraper since they all have different formats and you can’t rely on HTML structure.
I have published my initial attempt, and now trying to get serious on the parsing (the internals are far more powerful than the porcelain let it see). This is to be able to show errors like in https://docs.rs/ariadne/0.1.3/ariadne/ (yesterday finally the first!) and get some type inference/checking.
Plus, lossless parsing so it could work well with a editor.
This have proven to be harder than expected!
----
After it, I wish to work on the UI and embed a DB (sqlite) but also wish to build my own rdbms, how hard it could be? :)
I am building this because the Texture Atlas exporting feature from Animate CC is actually pretty awesome, yet only an Adobe Air runtime currently exists for it. I find Lottie restricted in its use-cases, can’t handle many animations, and I require raster images. My library can be used from loading spinners all the way up to entire games, with a simple but powerful model for manipulating animations dynamically at runtime.
[0] - https://derw-lang.github.io/
Github - https://github.com/eeue56/derw
Announcement post - https://derw.substack.com/p/why-derw-an-elm-like-language-th...
The current project in that direction is a timeline-maker targeting research labs. Imagine joining a research lab (for MS/PhD) and getting a timeline of key papers in their field. This will set a baseline for "literature survey". Some of my professors from my college have already shown interest.
It is nothing fancy though. A python script that take YAML input and produce an HTML output. Now producing that YAML input with all the data from 100+ papers is the tedious task. I am currently doing it for my field - speech recognition.
It started taking shape conceptually 5 years ago, but really started putting effort into it after the pandemic. It is such an intense pleasure to work on typefaces, and if I could, I would do this for the remainder of my life.
Latest update: https://neil.computer/notes/berkeley-mono-february-update/
If you're interested, please sign up here: https://berkeleygraphics.com/
Most people don't need new randos in their lives. They might have time to be in group chats with peeps of value. Selecting interests by unverified people isn't the solution, even though most apps go this route. Value + ease, is key to this. The are clever ways to group people and semi verify that have not been exploited yet.
What I am working on. If you are a skilled iOS dev or backend pro (cloud functions to be written!) and interested, DM. (or anyone else).
I'm still reading research papers to figure what's the best way to approach this problem. This is just the first step and I'll see what can be done regarding collecting data later on.
Reference: my dog has problems with his hips and likes to jump... Suddenly the pain kicks in.
Hope it helps
It turns Facebook Business Pages into Web Sites. Give it a try!
Neat trick, but it's a tough market to crack.
Fortunately, I have other side projects that need attention. :)
Next up is Headlamp. It's a Chrome plugin for web test UI coverage & test team monitoring/coordination. It's been offline and broken for years:
But ideas are cheap, a minimally viable product would be more impressive.
I mention Github because much like code, good content and knowledge should change and improve over time and have lots of commits and pull requests.
I mention Substack as users can subscribe to authors (they can also buy individual books).
Currently splitting my time between framework improvements in prep for a 1.0 (wrapping up a PostgreSQL integration today) and working on a deployment automation tool.
Because at this stage the startup has too much going on, this PaaS is very low overhead - so no Kubernetes complexity.
Think EC2/EBS/ELB at flat predictable rates.
It will be called inbox.jobs. No central website but each user will have its own xyz.inbox.jobs and xyz@inbox.jobs email
It's in crude form now, but I'm working to change that.
I have already implemented the garden lights and the heat on off. I get temp and humidity data from the garden and i want to add some security features and implement a morning routine (warmup the espresso machine, start the heat, open the curtain)
So far, the switch from jQuery, thrills me.
https://github.com/hofstadter-io/hof
Live streaming their development has been super fun too https://twitch.tv/dr_verm