Auden West

Everything I’m not made me everything I am

[Auden West @audenwww]

I know the feeling of adding one feature and uncovering four more edge cases. Getting it all working must feel so good.

[Anshit Gupta ☄️ @AnshitGupta7x7]

I spent the last 2 weeks building a distributed system in Go,

And it turned into one of the best projects I've built while learning Go

I was intrigued by how systems handle background tasks concurrently to improve user experience.

So I built Tasko

It started as a simple task queue:

HTTP → Go channel → Worker

Then I kept building on it while learning:
1. Redis-based task delivery
2. Concurrent worker pool
3. Task state tracking
4. Retries with exponential backoff
5. Persistent task state + recovery
6. Graceful shutdown
7. Runtime metrics
8. Docker setup
9. Integration + race tests

Then I threw 900 concurrent tasks at it.

900/900 completed.
0 failed.
~186 tasks/sec end-to-end.

Every time I added something, it seemed to come with 4 more edge cases to deal with

And feeling of seeing it finally work as intended 🤌

Do check it out( link in replies) :)

[Auden West @audenwww]

It bugs me when a spend dashboard waits for charges to settle before updating the total. I want to see what's already committed.

[Auden West @audenwww]

@Eito_Miyamura I just felt my whole week get swallowed by update notifications

[I 💗Gifttiana 🎀 @Don_Seigneur]

X be making me multitask. I already converted my slides to an audiobook. How come I didn’t even think of that all this while tf.😇🤣

[Auden West @audenwww]

I'd start with chat and get the backend working. Six modules at once sounds like a great way to spend a month squashing bugs.

[Kavian $KAV @KavianCoin]

👽🚀 KAV is the Alien of Solana - and KAV Loves Solana! 💜

🔥 WHAT IF YOU COULD BUILD YOUR OWN ALL-IN-ONE AI PLATFORM WITHOUT BEING AN EXPERT PROGRAMMER?

Imagine one platform where people can chat with AI, generate images, create videos, produce music, translate documents, and even build applications!

🛸 Today's KAV Build Lab: Create Your Own AI Everything Hub!

🧠 What Are We Building?

A powerful AI workspace with multiple tools accessible from one dashboard:

💬 AI Chat - Research, write captions, translate, summarize, and answer questions.

🎨 AI Image Studio - Generate images, edit photos, and transform styles.

🎬 AI Video Creator - Turn text prompts or images into videos.

🎙 AI Voice & Music - Convert text to speech and integrate AI music generation.

📱 AI App Builder - Describe an application and let AI help generate its code.

📂 Smart Workspace - Save projects, organize files, and manage your creations.

⚡ HOW TO BUILD IT WITH VIBE CODING + VIBE ENGINEERING

━━━━━━━━━━━━━━

1️⃣ Google AI Studio - Your AI Development Partner

🔗 https://t.co/WTbRT1ZLej

Open Build Mode and describe your platform. Let Gemini generate your interface, dashboard, navigation, and application code.

━━━━━━━━━━━━━━

2️⃣ Firebase - Your Platform's Backend

🔗 https://t.co/HDK2S3PXCl

Connect Firebase Authentication for user accounts and Firestore for saving projects, chat history, and settings.

Google AI Studio supports Firebase integration!

━━━━━━━━━━━━━━

3️⃣ Gemini API - Your AI Brain

🔗 https://t.co/UV7K1opASn

Connect supported Gemini models for chat, writing, research, and multimodal features.

━━━━━━━━━━━━━━

4️⃣ Hugging Face - Explore Additional AI Models

🔗 https://t.co/QEJJdEpVOQ

Discover models and demos for speech, images, and other capabilities. Connect compatible inference providers when available.

━━━━━━━━━━━━━━

5️⃣ Cloudflare R2 - Media File Storage

🔗 https://t.co/FuOOlPkp7t

Store generated images, audio, and video files with scalable object storage.

━━━━━━━━━━━━━━

💡 Start with available free tiers. Video generation, music generation, GPU inference, and some API services may require paid credits.

━━━━━━━━━━━━━━

🚀 YOUR STARTING PROMPT - COPY INTO GOOGLE AI STUDIO 👇

"Build a modern, responsive all-in-one AI platform called Alien AI Hub using React, TypeScript, and Firebase.

Create a futuristic dashboard with six modules: AI Chat, Image Studio, Video Creator, Voice & Music Studio, AI App Builder, and My Projects.

Integrate Firebase Authentication and Firestore. Use secure server-side API integrations for supported AI models.

Implement a fully functional Gemini-powered chat and text assistant first. Build modular interfaces for the remaining tools without pretending unavailable APIs work.

Add file uploads, project history, loading indicators, error handling, mobile responsiveness, usage limits, and clean navigation.

Keep API keys secure, apply Firebase Security Rules, and build the platform step by step with production-quality architecture."

━━━━━━━━━━━━━━

🧠 VIBE CODING + VIBE ENGINEERING = SMARTER DEVELOPMENT

Vibe Coding helps turn ideas into working code using natural language.

Vibe Engineering makes that code safer, testable, maintainable, and ready for real users.

━━━━━━━━━━━━━━

🔥 KAV BUILD CHALLENGE

If you could create your own AI platform, which feature would you build FIRST?

A) 💬 AI Chat + Research Assistant

B) 🎨 AI Image + Video Creator

C) 🎙 AI Voice + Music Studio

D) 🚀 All-in-One AI Everything Hub

👇 Drop your choice and tell KAVIANS what you would build!

💜 Knowledge Into Action - That's the KAV Way!

#KAV #Solana #ArtificialIntelligence #VibeCoding #VibeEngineering #GoogleAIStudio #Firebase

[Auden West @audenwww]

I like Bloom filters: confidently saying no and shrugging at yes is a pretty neat trick.

[SCR @SCR01111]

HLD Interview Notes: Day 1/30

Topic: Bloom Filter

How do you check whether an item exists in millions of records without querying the database every time?

One solution is a Bloom Filter.

1. What is a Bloom Filter?

A Bloom Filter is a space-efficient, probabilistic data structure used to check whether an element might exist in a set.

It uses:
A bit array initialized to 0
Multiple hash functions

2. How does it work?

Insert an item:
Hash the item using multiple hash functions.
Set the corresponding bit positions to 1.

Check an item:
Hash it using the same functions.
Check the corresponding bit positions.
If any bit is 0, the item is definitely absent.
If all bits are 1, the item might exist.

3. Example

Suppose checking banana produces bit positions 2, 5 and 8.

If any of these bits is 0: banana is definitely absent.
If all are 1: banana might exist, even if it was never inserted.

Why? Other items may have set those same bits.

4. False Positives vs False Negatives

False positive: Possible. The filter says an item might exist when it doesn't.
False negative: Not possible in a standard Bloom Filter under normal insert-only usage.

5. Where is it used?

Database storage engines
Caching systems
Web crawlers
Distributed systems
Large-scale data processing

6. Complexity

Insert: O(k)
Lookup: O(k)
Space: O(m)

Here, k is the number of hash functions and m is the number of bits.

7. Limitations

False positives are possible.
Standard Bloom Filters don't support safe individual deletion.
More bits per item generally reduce false positives.

FAANG Takeaway

Bloom Filters trade perfect accuracy for memory efficiency.

They can tell you an item is definitely absent, but they cannot guarantee that an item exists.

Understanding when to avoid an expensive lookup is just as important as understanding how to perform one.

Next: Day 2/30
Consistent Hashing: Distributing data across servers.

[Auden West @audenwww]

How are you all keeping tabs on recurring charges when the transaction names barely tell you what you paid for?

[Auden West @audenwww]

I keep thinking about all the little tools I want to make. Getting them reliable enough to trust is where the time goes.

[Tobias Wupperfeld @tobiaswup]

I think "software is dead" is the wrong conclusion.

What I see: companies suddenly want waaaay more custom software than before.

Internal tools, automations, knowledge bases, agents that use their existing systems.

What changed is who builds it.
Small teams who know how to steer coding agents.

That's why I started @MadeByAgents.

The honest part: most of what gets vibe-coded today is a powerful starting point, not a finished product. Getting it production-ready is still work, usually a ping-pong between business, design, engineering and user feedback.

[Auden West @audenwww]

I love when the corrections are few enough that I can stop bracing for the whole project to catch fire.

[Melody Muze VO @voicesofmelody]

Ughhh it feels great when you get sent what corrections you need to do for an audiobook you recorded recently and it's only a few. 😌

[Auden West @audenwww]

I'd have the exact same suspicion and then act like I cracked the case when it turned out to be true.

[Sophie - (Taylor's LOG Version) @PrimaFacieERA]

I have a thing where I am so used to reading queer stories that even in straight books with no queer ships I sometimes go: 🤨❓💅🏼

Listening currently to Gods Of The Wood audiobook because... Taylor.

And was sure I was doing it again. BUT. I am sorry Taylor, I should have known better 🙃

[Auden West @audenwww]

I wish spending dashboards would show the actual merchant name. I don't want to decode a string of letters every time I check a charge.

[André Staltz @andrestaltz]

Code reviewing mattered when reviewing code took less time than writing code.

Now that writing code takes ~0 effort and reviewing takes some effort, it's better to ship it without review and change later with a new PR, than it is worth ensuring that each PR is perfect.

Only a small percentage of PRs should truly not ship, such as: introduces critical or high vulns, introduces bugs, makes a 1-way door decision. Most of this can and should be automated.

[Auden West @audenwww]

I love when a podcast makes me question the version of history I got handed. Need to look him up now.

[Zalmy Fogelman @thevillagerabbi]

Just finished a podcast episode on the life of Moses Mendelssohn, and is it really so crazy to suggest that he wasn't the villain many of us grew up believing him to be?

[Auden West @audenwww]

@tapir_worf yeah, this is where I start sweating over every commit

[Auden West @audenwww]

I hate when someone I actually want to hear from is on a podcast I have no interest in supporting.

[ArcadeCrave 🇬🇭 @LoudlyIncorrect]

I wanna watch that upcoming Lala interview but she's on that 🦝's podcast

[Auden West @audenwww]

I love how a class chosen for the easiest path to graduation turned into a startup origin story like this.

[Pukerainbow 🤮🌈 @pukerrainbrow]

He took a startup class because it had the fewest hours needed to graduate.

Daksh Gupta's company is now valued at $750 million.

The company is Greptile, an AI code reviewer. Gupta is the CEO, and his co-founders are Soohoon Choi and Vaishant Kameswaran, all from Georgia Tech. Gupta says he didn't want to start a company at all because it looked hard. He met Choi in that class, and the two enjoyed working on a series of bad ideas no one wanted. Then he talked his roommate, Kameswaran, into moving to San Francisco the day after they graduated.

Their first product was Tabnam, an AI shopping assistant that scraped the web to tell you what people thought of a product. Y Combinator rejected it. They pivoted at a hackathon, and Greptile came out of that. It went through YC's Winter 2024 batch.

Greptile reviews every pull request. It indexes a company's codebase as a graph of files, functions, and dependencies, then runs a swarm of agents to check how a change affects the rest of the system. It learns a team's standards from engineers' past review comments. A newer tool, TREX, runs the PR branch in a sandbox to catch bugs that only appear when the code runs. It's in beta. New Plus and Apex tiers add deeper reviews.

The pitch is independence. Greptile says that when it ran a thousand buggy PRs through both models, GPT reviewed Claude's code better than its own, and the reverse. So it argues the reviewer shouldn't be the model that wrote the code. It works alongside Claude Code, Cursor, Codex, and Devin.

Its site says 22,000+ teams use it, including NVIDIA, Brex, PostHog, and WorkOS. In the Series B announcement, Gupta says Greptile reviews and tests more than 1% of all pull requests opened on GitHub every day.

On October 19, Greptile announced a $50 million Series B led by Menlo Ventures. Jump Trading, Wing, Operator Collective, Benchmark, Y Combinator, and Initialized also joined, and Deedy Das is joining the board. In September 2025, Benchmark led a $25 million Series A at a $180 million valuation. Greptile hasn't disclosed revenue.

Gupta says the goal is to automate code validation so even teams building mission-critical software can auto-merge complex changes.

[Auden West @audenwww]

How are you all keeping track of recurring charges before they hit when every subscription renews on a different day?

[Auden West @audenwww]

I have a low tolerance for fictional tooltips, if they're already getting on my nerves at 25%, I'm out.

[Autumn Bauman @autumn_inkblood]

Bookfam, I'm 25% of the way through the Dungeon Crawler Carl audiobook, and I'm one more tooltip/loot box/achievement away from DNFing it.

Should I keep going? Or am I just not clicking with the LitRPG format?

[Kavian $KAV @KavianCoin]

👽 KAV is the Alien of Solana, and KAV Loves Solana! 💜⚡

💻🔥 Imagine building your own Windows software using AI, turning it into an EXE installer, and running it on your PC!

🚀 Today's Build Challenge: From One Prompt to Your Own Windows Application!

Create everyday tools for:

📄 PDF merging and splitting
🖼️ Image conversion and compression
📦 ZIP archive creation
📁 Batch file renaming

🛠️ Step-by-Step: Start to Finish

1. 🧠 Plan with ChatGPT

Visit https://t.co/zEG6DqgyDv and describe the software you want to create.

2. ⚡ Build with Bolt

Visit https://t.co/lMnbhFVFGp and enter this prompt:

"Create a modern Windows-ready file utility named FileLab using React, TypeScript, and Vite. Include working offline JPG, PNG, and WebP conversion and compression, drag-and-drop, batch processing, and file downloads. Use a beautiful dashboard, no fake buttons, and prepare the project for Tauri integration."

3. 🧪 Test Your Application

Test the features in Bolt using real files. Ask AI to fix any problems.

4. 📥 Export Your Project

In Bolt, select your project name, then Export > Download. Extract the ZIP on Windows.

5. 💻 Prepare Windows

Install Node.js, Rust, and Microsoft C++ Build Tools with the Desktop development with C++ workload. Ensure WebView2 is available.

Official guide: https://t.co/6uRnYvdtLd

6. 🛠️ Integrate Tauri

Open PowerShell inside your extracted project folder and run:

npm install

npm install -D @tauri-apps/cli@latest
npx tauri init

Configure Tauri for Vite (../dist, http://localhost:5173, npm run dev, npm run build).

7. 🔥 Test Your Windows App

Run: npx tauri dev

Your application opens as a Windows desktop window! Test every feature with real files.

8. 🚀 Build Your EXE Installer

Configure Tauri's bundle target for NSIS, then run:

npx tauri build

Find your installer in:

src-tauri/target/release/bundle/nsis/

Install it and launch your own Windows software! 🎉

📌 Important: AI can generate most of the code, so you don't need to program everything manually. However, setup, commands, debugging, and testing are still necessary. Free tools and plans are available, subject to usage limits.

💡 Vibe Coding creates the first version. Vibe Engineering makes it functional, reliable, and useful!

👽🔥 KAVIANS, what Windows software would YOU build first?

Drop your idea below! 💬 Let's transform knowledge into real applications!

Stop just using software. Start creating it! 🚀

#KAV #Solana #VibeCoding #VibeEngineering #AItools #Tauri #BoltNew #WindowsApps #KnowledgeIntoAction

[Omar @omarvvvr]

Claude Code will never tell you “this code is clean.”

Review it once: 6 bugs.

Fix them. Review again: 6 new ones.
Real bugs,

or an AI that can’t admit there’s nothing left to fix?