My startup isn't "tech" enough so I hesitated before making this post, but I saw that people talked about their PhD or master thesis and all sorts of cool things, and after all, the prompt is "what are you working on", so here we go!

I am hard at work preparing the 2nd edition of Seoul SPARKS (seoulsparks.com), a 6 weeks pre-college summer program in Korea (currently aimed at US students but open to students all over the world). It is designed to be a high quality, rigorous but also as immersive as possible program, for high school students who want to deepen their understanding of Korean language, culture, and history while earning academic credit at Seoul National University. My aim is to educate the future generation of Korean Studies scholars and I hope my program can be a stepping stone for that! My program director designed a chronological history curriculum, that explores Korea’s past and present through site visits and expert-led discussions. Each week focuses on a specific era of Korean history, beginning with the Joseon Dynasty (1392–1897), where students visit palaces, wear hanbok, and learn about Confucian traditions. The following weeks cover the Japanese colonial period (1910–1945), the Korean War and its aftermath (1950s), Korea’s rapid economic rise through the Miracle on the Han River (1960s–1980s), and finally, modern Korea, including democratization, globalization, and the Hallyu wave. These themes are reinforced through guided tours of historical sites, museums, and cultural landmarks. We try to make it fun too and I think our students last year had a blast!

We try to offer a lot of other really cool things, such as a 100% Customized Cultural Activity (these have included K-pop dance training at a professional studio, Korean Sign Language classes, traditional calligraphy, culinary workshops, and even mini-internships with local businesses ), the opportunity to work on a research project under the supervision of a Korean PhD student if they'd like, and we also link them with university mentors (basically, Korean college students) who hang out with them, answer their questions about life in Korea, help them adjust to life here, and just have fun with them really.

Last year's pilot program was very successful, we were blessed to have a fantastic cohort of very bright and motivated students. The program was also highly rated by parents and students alike (the only serious negative comment we received was: why is Korea so hot and humid during summer?), and I hope we can continue to grow and keep attracting the best students from all over the globe!

Anyway applications are open until April 30th so if any of you have kids or know someone who could be interested, I would be very grateful if you could spread the word :)

Thank you for reading!

i'm improving my chat site aimed at programmers and technology lovers <https://chat-to.dev>

On the FOSS front, I've been doing some minor things in TXR Lisp land.

Here is one:

When the random/rand functions are called with small moduli, well under 32 bits wide, they still draw a 32 bit word from the PRNG. I fixed that. Now the random state has a 32 bit shift register which it can load from the PRNG and take bits from it (in batches of 2, 4, 8 or 16).

There is a way to do this without any additional state other than that 32 bit word to indicate that the register has run out of bits.

We pretend that the register is 33 bits wide, giving it an indicator high bit that is always 1. Thus whenever the value drops to 1 or 0, it has to be reloaded. We cannot fit that indicator bit into the register, so what we do is: when we are reloading it with a fresh value, we immediately take bits we need from that value, shift it right, which creates the space for the bit. We then mask the bit in the right spot, depending on the shift size.

I worked on enhancing brace expansion in the glob* function. The regular glob* function has brace expansion if it is provided by the POSIX C library one, but the extended glob* has its own. I added numeric and string range expansion to it. It's better than the corresponding features in bash, because you can use more than one character, as in {AA..ZZ}. Also, if you use leading zeros in numeric ranges you get leading zeros in the output, as in {000..999}. The step sizes are supported.

Inspired by the step sizes in brace expansion, I added the same to range iteration:

  2> [map cons 0..10..2 "a".."z"..3]
  ((0 . "a") (2 . "d") (4 . "g") (6 . "j") (8 . "m"))

... and other stuff. There are a number of items remaining on the TODO list marked for next release.

ever feel buried in applicants and wasting time on interviews that lead nowhere?i’m building an AI tool to help you instantly screen candidates so you only spend time on the best ones (and don’t miss out on great hires while stuck interviewing the wrong ones).

if that hits home, i’d love 30 mins to share what i’m building and get your thoughts. reply here or DM me if you’re up for a quick chat! Calendly link (feel free to lmk if any other times work better for you) - https://calendly.com/vibrane/30min

Very much WIP: I'm working on an alternative to subscriptions for online publications. Instead of subscribing to entire publications / blogs, publishers would register their publication on this network and configure thresholds and pricing. Add a bit of code to the site and a paywall will show up, allowing readers to pay for individual articles. The prices would be minimal, amounting to less than a dollar in most cases. i.e. reading articles using micro-transactions

I know it's been tried before, but I thought I'd attack it with a few different angles - web based, no chrome extension, and thresholds to help verify the article is worth it.

You can see the proof-of-concept here: https://paperwall.io/

zero-knowledge proof framework to prove for example that a bank has the necessary funds, or for a loan application that I make enough without revealing how much

building the quickest AI chrome extension (700k downloads): https://chatgptwriter.ai

A revolutionary product to disrupt sustainable agriculture.

lambdamoo rewrite https://github.com/rdaum/moor/

Writing an fft core in vhdl. Pipelining is not easy.