thomask1995

Shame, it was probably the best designed No-Code workflow builder out there

Very cool project!

I think some sort of gui to help write the cors headers would help tbh.

it's quite difficult to get your stack to work for local dev, CI, and prod since each most likely needs different cors headers. Especially if you use tunnels and proxies like we do.

What made it click for me though was understanding what problem it solved.

Def this right now

https://impeccable.style/

This is the app I was begging for a few years ago!

Ended up buying a ton of Ear Training books and using those little web apps.

This would have been so much better.

I just started using it,

I will say, I feel we can skip the chords annotations and stuff if the ICP is intermediate guitarist.

Or do a quick survey at the start and see what fundamentals they know.

For intermediate guitar, I think the main gaps are:

- Ear training (unless they were taught early) - triads + inversions - Music theory beyond basic pentatonic shapes - Synchronization at higher speeds (this was my biggest one by far) - Chords beyond basic ones and M/m 7th.

I feel you could add little quiz or survery to see if they are already familiar with some of these.

just my 2 cents

Overall, love this. Very happy someone took the time to do it.

This is super cool!

Have you thought about making a tool to help preview/dry run migrations?

I feel that's something I would want a ton of confidence in if an LLM is making migration scripts.

Especially if it's doing scary stuff like breaking up a table.

Very cool! Would be sweet to add number of usages each tool gets.

Interested in seeing what other people use!

HI! OG Author here.

Honestly, I don't know.

This was purely a toy project/thought experiment to challenge myself to learn exactly how these LLMs worked.

It was super cool to see the loss go down and it actually "train".

This is SUPER far from a the real deal. Maybe it could be cool to see how far a fully in memory LLM running on CPU can go.

License added! Good catch

I wanted to learn exactly how these things work. Down to every detail.

So I built a Mini LLM from absolute scratch in Rust. No libraries other than ndarray and rand.

Started a few months ago but got stuck. Used claud code and fixed the remaining issues and now it works!

I know I am about to get roasted. But hopefully someone suggests a killer tech-stack and I am proven wrong.

I really don't like coupling backend and front end, but at the same time, wanted extreme type safety on the REST API.

So I put together this very simple full stack TS boilerplate that syncs types via a shared folder.

Super lightweight, easy to extend, two seconds to set up.

Let me know what you think. Or roast me and suggest something better!

Very interesting. You have any before and after examples?

Curious how it handles some of the concurrency stuff. Actors, async/await etc..

OK I need to give this a go. tbh, I've been going back to just writing stuff manually and asking ChatGPT doc questions.

I talked to and extremely strong engineer yesterday who is basically doing exactly this.

Would love to see a video/graphic of this in action.

I found it hard to find content like this when I was looking to get promoted.

By far the best content came from Will Larson. Thought I would add another data point here to help whoever is currently looking to get promoted.

(ps: yes I wrote this a few months ago, but a friend said I should share here for more vis)

I find they are much better at tailwind css. Been really impressed lately actually, and I was a skeptic

Cool idea!

I see this does consumption, and plans on output?

I imagine that's a lot harder as LLMs are training on traditional MD.

Overall, I think there is a pain point here with MD + LLMs. I find myself having to beg is to restrict what Markdown to use because a lot of places only support a subset (ex: SwiftUI)

Completely fair!

I was considering making this CLI based and running locally.

But the design partner I had preferred web based with the feed.

If there is enough demand, I will gladly make a local version.

It would just need an OpenAI Key + set up an alias or pre-push script to intercept git push.

Great point!

Yea right now, it doesn't do any de-duping. For sure a feature I can add.

I think there is a whole other side to this where we make PRD -> Tickets streamlined as well as a Backlog grooming assistant.

For sure need to give it more thought though.

I honestly think this could help you build a following on X lol

I maintain a search index with all your Tickets.

Then when a commit comes in, I run a semantic search to pull up candidate tickets.

I then feed that to the prompt, and let the AI decide. The challenge is the index for sure, those params definitely still need tweaking.

Hey!

Yea I am half way to a working Jira integration. Turns out, it works super differently than linear.

And yes! Talking to the AI in Slack is for sure a feature I want to support.

Just want to make sure I build the right thing. So sharing early