
Building @clankeroverflow 🇲🇦 | Programmer | Builder | Distilling knowledge from humans and LLMs
127.0.0.1
Joined December 2021
- Tweets5.2K
- Following3.1K
- Followers1.2K
- Likes68.6K
we got humanoids before GTA 6
A true super-humanoid hero
I first saw Neo1 at WRC 2026 and was genuinely impressed. The 4-meter, 1.2-ton humanoid from Beijing-based NeodraDynamics has returned at CIIF 2026.
Neo1 is built for 4D work-- heavy assembly, deep-sea operations, port and mine material handling, nuclear facility maintenance and fuel transfer, emergency rescue, and high-altitude construction...These are dangerous, dirty, high-intensity jobs it is designed to take on in place of people.
It uses a hydraulic-electric hybrid drive with 41 active DoF-- 39 hydraulically actuated joints and two electrically driven forearm rotations.
Each arm is rated for a 300 kg payload, with a combined two-arm payload of 500 kg. Its dexterous hands claim 500 kgf of grip force and 0.01 mm fingertip precision, while the arms have a 1.7-meter working radius.
Neo1 is also designed for extreme environments, with claimed operating temperatures from −50°C to 250°C and depths of up to 10,000 meters, alongside tolerance to radiation, dust and vibration.
NeodraDynamics is bringing Neo1 into JD Industrial’s supply chain and industrial sites to test how it performs in real-world applications.
>Be polymarket
>pick some old article/video
>post about it on X and slap "JUST IN: " at the start so it sounds like breaking news
JUST IN: Chinese bride surprises her groom with a $12,000 Nvidia RTX 5090 GPU at their wedding, bringing him to tears.
Readers added context they thought people might want to know
The video and photos are from a 2025 Chinese wedding (not a recent "JUST IN" event). The ASUS ROG Astral RTX 5090 was reported as ~$3000 at the time per launch pricing and contemporary coverage, not $12000.
tomshardware.com/pc-components/…
videocardz.com/newz/asus-reve…
en.gamegpu.com/news/zhelezo/v…
Oussama retweeted
Introducing overshoot, it will cut your GPU bill in half with zero quality loss, or you don’t pay.
Over the past few months, we have forward deployed into AI teams to optimize their inference stacks. They have now cumulatively saved over $5M.
Overshoot is software that bakes in our hard-earned inference learnings into a scalable solution.
overshoot.ai
Oussama retweeted
Help us bring more awesome games to Linux
This is an open call to Linux distro maintainers and the wider Linux community, with a goal to work together and bring fair multiplayer gaming to Linux players.
We’re rebuilding our Linux anti-cheat into EAC Linux v2 so that Linux players can have a similar level of protection against cheaters as Windows users do.
We’ll pilot EAC Linux v2 with Fortnite first, with wider availability for game developers once we’ve validated its effectiveness.
Learn more and get in touch:
onlineservices.epicgames.com…
Oussama retweeted
اليوم عندنا بزاف ديال Open-Source Models اللي كيدعمو الدارجة والعربية بصفة عامة 🇲🇦
ولكن واش كلشي كيعرف كيفاش يخدم بهاد الموديلات؟ وواش عندنا Tools اللي كتسهّل علينا نستافدو منهم فـ real projects؟
هادي واحد من الأسئلة اللي خلاتني نبني Klam...🧵 👇
Oussama retweeted
I built an inference server from scratch in C++ to answer one question:
what does dynamic batching really buy you on a CPU?
It's called finton. Hand-written scheduler, worker pool and HTTP server, ONNX Runtime underneath.
Here is what I measured 🧵
> i noticed the pairing is slow
> no problem, we'll add a long animation during that, so you won't notice
Did you know about this?

























