code_bucks's profile picture. Software Engineer who ships AI apps | 35K+ YouTube subs | Scaling AI first development Agency | DM for AI SaaS development

CodeBucks⚡

@code_bucks

Software Engineer who ships AI apps | 35K+ YouTube subs | Scaling AI first development Agency | DM for AI SaaS development

고정된 트윗

What used to take designers hours now takes 30 seconds...⚡ 🚨 BREAKING: AIdea Central just went LIVE and we're about to democratize AI for everyone! 🔥 10+ tools in just one app! The magic we just unleashed: 🎨 Professional product visuals in clicks 💸 Create video ads of…

code_bucks's tweet image. What used to take designers hours now takes 30 seconds...⚡
 
🚨 BREAKING: AIdea Central just went LIVE and we're about to democratize AI for everyone! 🔥

10+ tools in just one app!

The magic we just unleashed:
🎨 Professional product visuals in clicks
💸 Create video ads of…

I've played around with multiple LLMs, and I'm starting to think we need a whole new approach to get to AGI. Honestly, I'm a fan of how LLMs have made boring day to day tasks easier. But yeah, I know they're not ready for complex coding or cutting-edge stuff yet.

(Low quality opinion post / feel free to skip) Now that AGI isn't cool anymore, I'd like to register the opposing position. - AGI is coming in 2026, more likely than not - LLMs are big memorization/interpolation machines, incapable of doing scientific discoveries and working…



Great post! 👏 It's a good read if you wanna figure out which tool to use for your programming tasks, like Codex, Claude Code, or cursor.

📢 Time for an update on my workflow. This one's a 23 min read, so buckle up. 100% organic and hand-written, like an animal. steipete.me/posts/just-tal…



CodeBucks⚡ 님이 재게시함

Excited to release new repo: nanochat! (it's among the most unhinged I've written). Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,…

karpathy's tweet image. Excited to release new repo: nanochat!
(it's among the most unhinged I've written).

Unlike my earlier similar repo nanoGPT which only covered pretraining, nanochat is a minimal, from scratch, full-stack training/inference pipeline of a simple ChatGPT clone in a single,…

CodeBucks⚡ 님이 재게시함

"Don’t do things that you know are morally wrong. Not because someone is watching, but because you are. Self-esteem is just the reputation that you have with yourself. You’ll always know." @naval


This is interesting👇🏻 Radical Numerics launched RND1, the most powerful open-source diffusion language model yet! This 30B parameter model (with just 3B active via sparse MoE) will change how AI handles reasoning and code, all while being fully open. How it is different from…

Introducing RND1, the most powerful base diffusion language model (DLM) to date. RND1 (Radical Numerics Diffusion) is an experimental DLM with 30B params (3B active) with a sparse MoE architecture. We are making it open source, releasing weights, training details, and code to…



CodeBucks⚡ 님이 재게시함

As of today, @shadcn is the most popular design system on @github ⭐, surpassing Material UI and Ant Design. Its authorship remains a grand mystery.

rauchg's tweet image. As of today, @shadcn is the most popular design system on @github ⭐, surpassing Material UI and Ant Design. Its authorship remains a grand mystery.

Looking forward to posts like: - It's over for n8n and zapier - It's game changer for automation - now you can automate your whole life from openAI agent builder💀


Really wish @cursor_ai had this: → Pick one model for planning → Another for execution Imagine GPT-5 or Claude Sonnet 4.5 handling the big-picture thinking, then Grok code fast or Kimi k2 jumping in to actually ship it. feels like the dream setup 👀


🚨Just saw @shadcn dropped these new components in shadcn/ui 🔥 They aren't flashy hero sections but they'll make your product feel very polished. Here is the quick summary: → Button group This is like a smart organizer for your app's buttons. Instead of buttons floating…

code_bucks's tweet image. 🚨Just saw @shadcn  dropped these new components in shadcn/ui 🔥

They aren't flashy hero sections but they'll make your product feel very polished.

Here is the quick summary:

→ Button group
This is like a smart organizer for your app's buttons. Instead of buttons floating…

CodeBucks⚡ 님이 재게시함

AI is intelligent to the extent it can get you what you want. AGI is intelligent to the extent it can get itself what it wants.


IDE >>>> CLI unless you are vibe coding!


CodeBucks⚡ 님이 재게시함

This is wonderful and I'd love to see more companies embed their taste into writing. It's good for both humans and agents. vercel.com/design/guideli…


10 Next.js project ideas for beginners💪

code_bucks's tweet image. 10 Next.js project ideas for beginners💪

Can anyone suggest a good keyboard? Not a fan of heavy mechanical ones or loud clicks, but a light, minimal mechanical feel + soft sound would be perfect🤍 Which ones are you using?


In next few months, your IDE agent will: 🔸plan → run → verify across your repo 🔸attach a test plan to every PR by default 🔸route models per file type (docs vs tests vs build) 🔸run in an ephemeral sandbox before proposing diffs 🔸log every decision for compliance 🔸prefer…


CodeBucks⚡ 님이 재게시함

Your database is slow? Try these 4 fixes first (before adding more servers): 1️⃣Add indexes on columns you query frequently 2️⃣Limit SELECT queries, don't SELECT * instead specify columns 3️⃣Use EXPLAIN to see query execution plan 4️⃣Paginate results, never return more than 100…

code_bucks's tweet image. Your database is slow? 

Try these 4 fixes first (before adding more servers):

1️⃣Add indexes on columns you query frequently
2️⃣Limit SELECT queries,  don't SELECT * instead specify columns
3️⃣Use EXPLAIN to see query execution plan
4️⃣Paginate results, never return more than 100…

This month I used 93M+ tokens in Cursor 📈 Surprisingly, I ended up using Grok code fast the most. it's insanely quick and nails tool-calling with scary accuracy. You just need to provide the right set of prompts.

code_bucks's tweet image. This month I used 93M+ tokens in Cursor 📈

Surprisingly, I ended up using Grok code fast the most. it's insanely quick and nails tool-calling with scary accuracy. 

You just need to provide the right set of prompts.

Loading...

Something went wrong.


Something went wrong.