scientaster2
scientaster2 on Hacker News
One nice part is the ability to run `-h` and get all of the commands/args. I wish that existed for rest endpoints too :)
Any bets on how long it'll take for a security breach, now that every attacker knows affirm is vibe coding 60% of PRs?
I feel like these top down mandates miss the forest through the trees -- in isolation claude code is a speedup, like how sometimes WD40 is the right tool for the job. But when applying it to everything, you end up with a sticky mess.
Alright then, post your ML simulation of a piano player, I'll wait :)
If it helps, some of the most successful people I know made their achievements by taking a risk on their current capabilities. It's obvious to say, but the best way to grow is to be in a position where you have a good growth opportunity - and inherently, that means you'll be swimming in a slightly bigger pond than you're comfortable in.
I don't know anything about the specifics of your situation, so definitely discount this advice if it's not applicable - but remember that every high level / successful person had to be, at some point, wholly out of their depth in order to rise to the occasion!
> Founded in 2016, Neuralink has devised a sewing machine-like device capable of implanting ultra-thin threads inside the brain.
I know people get elective surgeries all the time (plastics, etc), but I feel hard pressed that this company is going to convince people to open up their skull for some promise of cool technology.
Even if it is successful from a technology perspective, the chicken and egg problem they'll face with developers is an order of magnitude or two higher than with AR/VR headsets. It's already an uphill battle without having to convince people to open up their brain, I just can't see this gaining any adoption unless there's a surgery-lite path outside the skull.
The first time I visited NYC, I was appalled that so many people just casually live in the stink of street trash 24/7. They just don't see it as an issue, strangely
Posted by
scientaster2 on Tuesday October 17, 2023 @12:56PM
from the ask-hn dept.
I'm looking for my next side project and wanted to get some feedback from anyone working on LLM-Powered apps. I'd love to build some kind of tool to assist with the process - I just need to know what problems you're having!If there's something that annoys you daily, or some platform you're surprised doesn't exist yet for teams building in the AI space, it'd be excellent if you could share it...
Posted by
scientaster2 on Tuesday October 17, 2023 @08:05AM
from the self-post dept.
Hi folks! I'm looking for my next side project and wanted to get some feedback from anyone working on LLM-Powered apps. I'd love to build some kind of tool to assist with the process - I just need to know what problems you're having!If there's something that annoys you daily, or some platform you're surprised doesn't exist yet for teams building in the AI space, I'm all ears!
I'm building a GraphQL based headless CMS currently, ~2 months away from an official launch. It's currently wholly lacking official documentation but feel free to give the index page a peek and see if it's something you'd be interested in trying out!
(or if you just have general feedback or learnings from your own experience, I'd love to hear that too!)
https://brick-cms.com/
Thoughts on this?
> 4. Don’t choose a large market
I feel like there's also a case where the opposite is true, e.g. that there's already proof that businesses are willing to pay for solutions in the space. As long as you have a differentiating factor to the big players (ideally via something they couldn't easily shift to), my knee jerk reaction is that it's still not a bad call to launch in the space.
Sci-fi extrapolation: I like the idea of there existing a few super smart LLM's trained on raw human data that function like generals, 2nd-order smaller models that are like lieutenants, eventually down to AI entirely trained on AI -- grunts that lack a lot of context but achieve a single purpose through language.
I've given server components with Next.js an honest try in my personal projects, and honestly, the pain of using them far outweighs the benefits.
I MUCH prefer the pattern of getServerSideProps seeding data on-page-load and working with it via context providers. The fact that data loaded in with server components cannot support reactivity of any kind means you have to litter your code with "use client" wrappers that take in their parent data as initial state, then introducing client reactivity on top.
That^ also means that any time you drop into "use client" you drop any SSR benefits anyway. I agree with a lot of this article that the transition feels forced, and I can't help but hope that Vercel commits to supporting pages style routing indefinitely.
Here's an example, the pattern makes it almost impossible to highlight a currently active route in a navbar. Literally a basic building block of web applications: https://github.com/vercel/next.js/issues/43704
Posted by
scientaster2 on Wednesday May 24, 2023 @12:34PM
from the show-hn dept.
Posted by
scientaster2 on Tuesday May 23, 2023 @09:03AM
from the show-hn dept.
https://github.com/Steven-Ireland/spam
There are plenty of good tools for load testing such as JMeter, Minigun, and more - but they all have their own set of options to learn.
I built this one to support a simpler workflow where you just paste your curl request (or other cli command) after your `spam` config. literally just:spam -r 2 -- curl www.google.comIt's pretty barebones but I'm 100% going to use it in my day to day - figured...
Posted by
scientaster2 on Sunday May 21, 2023 @03:32PM
from the show-hn dept.
https://github.com/Steven-Ireland/spam
There are plenty of good tools for load testing such as JMeter, Minigun, and plenty more - but they all have their own set of options to learn.I built this one to support a simpler workflow where you just paste your curl request (or other cli command) after your `spam` config (literally just `spam -r 2 -- curl www.google.com).It's pretty barebones but I'm 100% going to use it in my day to day -...
I switched to decaf and it's really the best of both worlds - water processed beans have a great taste without the caffeine crush
Remember though that as you approach the speed of light, relativity starts to kick in with your time "slowing down". In other words, the inverse relationship between speed and time could allow civilizations to expand faster than it would seem possible at "normal time" speeds. Complete conjecture on my part, just a fun thing to think about!
Posted by
scientaster2 on Monday February 08, 2021 @10:06AM
from the show-hn dept.
+1 to remote! As an east coaster, I've participated in many hackathons in person and I can vouch that none have actually benefited from colocation. Since teams are competing, there is naturally no incentive to associate with peers outside your direct team - which you can do plenty well remotely.