Puneeth G R

Software Engineer

[Unsloth AI @UnslothAI]

Introducing Unsloth for AMD 🚀
You can now train & run LLMs on your AMD hardware

• We collaborated with AMD to enable you to train & run 500+ models on AMD GPUs
• Works on Windows, WSL, Linux
• Train Qwen, Gemma on 3GB VRAM

GitHub: https://t.co/2kXqhhvLsb

Works on Radeon, Instinct, Ryzen and data center GPUs with up to 2× faster with 70% less VRAM and no accuracy loss via our custom Triton kernels and math algorithms. We also support optimized ROCm builds for GGUF & Safetensors inference.

Unsloth is an open-source local UI for faster LLM training and inference, with tool-call healing, code execution, secure web search, remote APIs, and HTTPS deployment. Connect local models to Claude Code, Codex agents and run the latest Kimi, GLM, DeepSeek, Qwen3.6, and Gemma 4 models.

🔗Blog + Guide: https://t.co/U9LqyRjFdj

[Milan Jovanović @mjovanovictech]

𝗪𝗵𝗮𝘁 𝗶𝘀 𝗮 𝗪𝗲𝗯𝗵𝗼𝗼𝗸?

Here's an example to help you understand it.

You have an application, and you want it to automatically do things without you having to trigger them manually.

That's where webhooks come in!

Webhooks are like messengers, allowing one system to send a message to another whenever a specific event happens. It's like getting a notification on your phone when someone comments on your social media post or when you receive an email.

Webhooks work similarly but for systems and applications.

Here's the basic flow:

1) Service A exposes a webhook
2) Service B subscribes to the webhook from Service A
3) Service A writes down the subscription
4) The event occurs in Service A
5) Service A sends the webhook request to Service B

There are many ways that webhooks can be helpful.

For example, you can use webhooks to automatically update a database or trigger an action when a form is submitted on your website.

They can also be used to send notifications or alerts, such as when a new user signs up for your service or when an error occurs in your system.

I once implemented a webhook in a text-to-speech API.

The application exposes an endpoint to upload a text file, and you can also pass in a webhook URL. An AI model will process the text file in the background and convert it into an audio file. When the audio file is complete, the application will trigger the webhook. The client can then download the audio file or do something else.

Did you have a chance to build something cool with Webhooks?

[Raul Junco @RaulJuncoV]

I’ve seen the Dual Write Problem break production systems.

The classic example:

1. Save data to a database
2. Publish an event
3. The application crashes between both operations

Now the database changed, but the event never reached the broker.

Two common solutions:

Change Data Capture

CDC reads committed database changes and publishes them as events.

Trade-off: less application logic, but more operational complexity.

Outbox Pattern

Store the data change and event in the same database transaction. A background process publishes the event later.

Trade-off: stronger application control, but you must manage the outbox and publisher.

Neither creates instant consistency across systems. They give you reliable eventual consistency.

What other solutions have you used?

[SecurityWeek @SecurityWeek]

Unpatched Claude for Chrome Flaw Lets Extensions Read Gmail, Calendar https://t.co/1dscu7icAR

[Docker @Docker]

What actually changes when you run an agent inside a Docker Sandbox?

This hands-on guide by @ajeetsraina walks through the setup, the isolation model, and the core concepts behind Docker Sandboxes. https://t.co/iwfMaLtZ48

[Puneeth G R @PuneethGR7]

Lite XL is a free and open source (FOSS) alternative to Sublime Text: https://t.co/OmbZ96qKGL

It's based on Lua programming language and it's available for Linux, MacOS and Windows. LiteXL has support for extra plugins too.

[Harvard Business Review @HarvardBiz]

A person can have first-class training, an incisive mind, and an endless supply of good ideas, but they won't be a great leader without emotional intelligence. https://t.co/9eJvhAI5yc

[Raul Junco @RaulJuncoV]

Most engineers overcomplicate locking.

Here’s the truth:

Optimistic Lock → Application-level control
Pessimistic Lock → Database-level control

Optimistic = let everyone edit, check for conflicts on save.
(Think Google Docs.)

Pessimistic = block everyone else until you’re done.
(Think library book.)

Rule of thumb

If retry is cheap → use Optimistic
If retry is expensive → use Pessimistic

[Fernando @Franc0Fernand0]

Tested code is not always good. But good code is always tested.

The testing itself doesn’t make the code better. Careful developers do.

The same care that produces clean code produces clean tests. Sloppy tests slowly drag the production code down to their level.

Code carefully.

Test it well.

Push coverage as far as practical.

And remember: not every test has to be an automated unit test.

[The Register @TheRegister]

Irish datacenters now guzzle 23% of the country's electricity https://t.co/vGR1TVhZC4

[Cindy Sridharan @copyconstruct]

Pre AI, people wrote poor commit messages, but you could still read a bunch of them and generally know what people were trying to do.

AI writes commits so verbose I’m in the “I ain’t reading all that. I’m happy for you tho, or sorry that happened” territory.

“More” ≠ “better”

[Ivan Velichko @iximiuz]

https://t.co/62EVb4GTLy

[WSJ Business News @WSJbusiness]

A coalition of record label and artist groups is pushing streaming giants to label music generated with AI, as the industry grapples with how AI is changing the business https://t.co/G7VYiCO2SR

[Cloudflare @Cloudflare]

We're opening the waitlist for our Monetization Gateway, which will allow you to charge for any web page, dataset, API, or MCP tool behind Cloudflare. The charges will settle in stablecoins over the x402 open protocol. https://t.co/pvICtEIixj

[TED Talks @TEDTalks]

7 ways to practice emotional first aid: https://t.co/BFeoiglkUj

[Alex Xu @alexxubyte]

Pagination Strategies for Large Systems

[Harvard Business Review @HarvardBiz]

If people are comfortable enough to speak their minds, conversations get broader and deeper. https://t.co/LJ4X7MtFYI

[Harvard Business Review @HarvardBiz]

If you don't apply information you learn, you'll forget 75% of it after just six days. https://t.co/8ypRGbYHvh

[Cloudflare @Cloudflare]

DNSSEC failure broke the .de TLD, risking millions of German domains. See how 1.1.1.1 used "serve stale" and Negative Trust Anchors to restore resolution during the incident. Read the full technical breakdown.

https://t.co/FtyON7H3QQ

[Ivan Velichko @iximiuz]

If you wrote a blog post or two, you likely noticed that your understanding of the topic _before_ and _after_ writing about it differs 1:10.

The same is true about reading tutorials vs. actually building or fixing stuff. You gotta be actively doing things to learn and grow.