Unclecode (Hossein)
@unclecode
Followers
3K
Following
3K
Media
270
Statuses
2K
Author of Crawl4AI (#1 GitHub Trending). Founder of Kidocode, SE Asia's largest tech & biz school. Leading AI research on synthetic data, AI consultant.
Singapore
Joined September 2009
🎉 Crawl4AI 0.7.7 is live, Self Hosting is the KING and QUEEN 🤴 🫅 Self hosting just leveled up. This release turns @crawl4ai into a real platform, not a library. You get a monitoring dashboard baked into Docker, open slash dashboard and you see the entire system breathing in
github.com
🚀🤖 Crawl4AI: Open-source LLM Friendly Web Crawler & Scraper. Don't be shy, join here: https://discord.gg/jP8KfhDhyN - unclecode/crawl4ai
2
5
30
Introducing 1st full-stack AI Designer An AI agent that can plan, generate graphic assets, Video animation, production UI code all with full context Design like this could easily cost $5~$10k just 1 year ago, but now is possible to deliver in 10 min without any design skills
13
49
430
This open the door to many domain specific small language models, I hope more people adopt such things to less distract by LLM and more focus on SLM
We managed to get Claude code, Codex and Gemini CLI to train good AI models thanks to @huggingface skills and you can too even (especially?) if you've never trained a model before 🤯🤯🤯 After changing the way we build software, AI might start to change the way we build AI
0
2
3
No reason to go to Upwork to hire designers anymore Here I just finished 2 jobs within one min with AI designer
18
6
179
Tried this on @SuperDesignDev "A minimalise character, head only, floating gently on a clean light off-white background. The character is a diffused, glowing sphere of flowing aurora light, blending cyan, violet, and soft blue gradients, with soft, foggy edges blending into the
8
10
204
This is how I prefer my AI assistant to be! I wish it had "None" option for all "Less" here in the list
0
0
2
Introducing Component Grab @SuperDesignDev Chrome extension allows you to grab any UI component in browser We turn messy DOM into clean tailwind code pixel perfectly Paste into Superdesign AI native canvas to vibe design Or paste into Claude code/Cursor Free forever:
This is the future of AI design/prototyping 1/ Import real product context 2/ Give Design agent a brief 3/ Unlimited design immediately, all on-brand and in-context And also... 👇🧵
33
60
692
Interesting take on eval by @eugeneyan Spend 4 weeks building a human-labeled dataset of 200 samples. Then prompt-engineer (crafting, learning, call it whatever you feel better) an LLM judge that aligns with that human data. Once aligned, run evals on your app prompt 100 times a
After repeating myself for the nth time on how to build product evals, I figured I should write it down. It's just three basic steps(i) labeling a small dataset, (ii) aligning LLM evaluators, and (iii) running the eval harness with each config change. https://t.co/HjUL3yZQPk
0
0
2
📞🤖 Phone Calling Agents Course Made by the LangChain Community Build production-ready voice AI agents for real calls via Twilio. Uses Opik (native LangChain integration), open-source models, multi-week course with live sessions. Check it out: https://t.co/DdRosZMkiV
10
90
541
Programmatic Tool Calling, this more important than opus 4.5! That is what I always say and do: An agent is just llm with only one capability, “code execution”! Let that code call other tools, api, mcp… Now Claude has it nicely
Alongside the model, today we're launching three very useful API features for building agents that scale to hundreds of tools without context bloat. - Tool Search Tool - Programmatic Tool Calling - Tool Use Examples Here's how they work:
0
0
2
This is big, go try it out!
🎉 Crawl4AI 0.7.7 is live, Self Hosting is the KING and QUEEN 🤴 🫅 Self hosting just leveled up. This release turns @crawl4ai into a real platform, not a library. You get a monitoring dashboard baked into Docker, open slash dashboard and you see the entire system breathing in
1
1
2
Crawl4AI Cloud now runs only ~7% slower than local, under 100 concurrent requests. The challenge isn’t speed, it’s scaling without losing what makes it Crawl4AI. Local feels personal. Cloud feels distant. I’m trying to erase that line.
I’ve been running some internal benchmarks testing the upcoming Crawl4AI Cloud against the local engine. The challenge: make the cloud feel as fast and efficient as the local version on my MacBook Air. Local still leads by a small margin, but the gap is closing fast. Once both
0
2
6
A transcription tool that finally transcribes faster than you think. I’ve used many, Willow Voice, SuperWhisper, even built my own, but this new one is insanely fast, beautifully designed, and made by a talented friend. Brand new, may have minor quirks, but worth trying.
whryte.com
Turn speech into text anywhere you can type, completely offline. 100% private, optimized for Apple Silicon.
I am launching whryte to help you type faster with your voice. - one of the fastest on-device voice transcription app for Mac - 100% local and private - no subscriptions - under 100MB for transcription model (without LLM)! - enhance mode (STT+LLM) - smart auto-edits. - highly
0
0
4
I’ve been running some internal benchmarks testing the upcoming Crawl4AI Cloud against the local engine. The challenge: make the cloud feel as fast and efficient as the local version on my MacBook Air. Local still leads by a small margin, but the gap is closing fast. Once both
2
1
7
Just crawled and converted the 4.5 MB (~300 pages) Smol Training Playbook from Hugging Face into clean markdown in ≈10 seconds, using @crawl4ai, all locally on my MacBook Air. (watch the video) Two major commercial crawlers on the same page: one took ~40 s, the other timed out!
3
3
23
A good template for anyone wanna build an agent that works just good
💻Deep Agents CLI Using the deepagents package, we built a simple coding CLI as an example of a coding application you could build on top of deepagents We added in a concept of memory so that it would remember instructions and guidance over time Blog: https://t.co/aX2jRKwuh8
0
2
13
I am not unemployed and no need to be and hell yea will read this book 💪💪
Training LLMs end to end is hard. Very excited to share our new blog (book?) that cover the full pipeline: pre-training, post-training and infra. 200+ pages of what worked, what didn’t, and how to make it run reliably https://t.co/iN2JtWhn23
0
1
0