Codex AI Korea Community Leader on Facebook, HPE Sr. HPC&AI PreSales Solution Architect, Former Microsoft Sr. Technical Evangelist,
대한민국 서울
Joined January 2009
- Tweets8.8K
- Following3K
- Followers3.4K
- Likes3.5K
Pinned Tweet
Hi Friends, I’ve put together a recap of Albert Yip’s technical session at OpenAI DevDay 2026 Seoul last week. From OpenAI dots and Specialist dots to the Decisions API and Agents API, I was impressed by how AI makes decisions and carries out work. For enterprise adoption, it’s important to define which tasks to delegate, what access to grant, and when human approval is required. Which repetitive tasks on your team could you hand over to ChatGPT?
synabreu.github.io/openai/%E…
#OpenAI #DevDay #AIAgents #WorkflowAutomation #ChatGPT
synabreu retweeted
GPT-6.1 Terra: delivering fast and capable computer use at a lower price.
GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.
🤖 Made with AI
synabreu retweeted
We’re excited to announce our new CapCut plugin is coming soon to @ChatGPT at CapCutWorldLA.
The plugin brings CapCut's video editing and generation tools alongside ChatGPT. You can start with an idea in ChatGPT, bring it to life with the CapCut plugin right inside ChatGPT, then take your draft video to CapCut's web editor to refine, share and publish.
By bringing the CapCut product you already love and use in ChatGPT, we're making content creation more seamless and efficient.
synabreu retweeted
Want to know what people are building with hardware and Codex?
🧵 Here are cool projects to take inspiration from:
xitter.cf/covacut/status/2074330…
synabreu retweeted
إنفيديا تطلق أضخم مَورِد مجاني في تاريخ الذكاء الاصطناعي والروبوتات! 🚨
أطلقت NVIDIA للتو أكثر من 2,575 مجسماً ثلاثي الأبعاد جاهزاً للمحاكاة فوراً، بملف ضخم حجمه 472 جيجابايت متاح مجاناً على منصة Hugging Face! 📂
المجموعة المسماة NVIDIA SimReady Assets ليست مجرد مجسمات عادية، بل تشتمل على خصائص فيزيائية واقعية وتسميات دلالية دقيقة تُمكن الذكاء الاصطناعي من فهم الأبعاد والكتلة والبيئة المحيطة!
كيف تُمكّن هذه الثورة مهندسي المستقبل؟
• 🤖 تدريب الروبوتات: تعليم الأذرع الآلية التقاط الأجسام وتحريكها بدقة متناهية.
• 🏭 بناء المصانع الافتراضية: إنشاء مستودعات كاملة بنظام «التوأم الرقمي» (Digital Twins).
• 📸 توليد بيانات التدريب: إنشاء صور وبيانات معنونة جاهزة لتدريب نماذج الرؤية الحاسوبية.
• 🎨 تأثيث العالم الإفتراضي: بناء بيئات 3D غنية ومعقدة بضغطة زر.
خطوة جبارة تقربنا أكثر من عصر الروبوتات الذكية والمصانع الذاتية بالكامل!
رابط التحميل المباشر للملفات في أول تعليق 👇
synabreu retweeted
Wonderful day at the track! Congratulations to Kimi Antonelli and the entire @MercedesAMGF1 team on a fantastic victory at #AustinGP! 👏
synabreu retweeted
On Saturday before @OpenAI DevDay there was DevDay Exchange Community Hackathon Berlin hosted by @maxjendrall and @paw_lean.
Over 250 Builders came together to use Codex to “just build things”. It was amazing to meet so many great people and it was a pleasure to host the mini-workshops. Thanks @GreeChivet, @mrlesk, and @artur_wala for giving spontaneous talks!
Also happy to finally meet @reach_vb and @paw_lean from OpenAI and the amazing Ambassadors in person: @mrlesk ,@artur_wala , @laraavci_ , @janpfrenger and of course host @maxjendrall.
Thanks everyone that joined! Looking forward to DevDay today!
synabreu retweeted
#NVIDIAGTC Berlin: 젠슨 황 CEO의 키노트에서 공개된 NVIDIA의 AI 미래 전략을 한눈에 알아보세요! 🌍✨
유럽 기술 생태계에 약 100억 달러를 투자하며 AI의 저변을 확장하는 NVIDIA의 비전을 블로그에서 자세히 읽어보세요 🔗 nvda.ws/4snD5Ax
🏭 AI 팩토리: AI 팩토리 구축 기간을 획기적으로 단축하는 DSX POD 공개
🤖 AI 에이전트: OpenShell Reference Architecture for Safe AI Agents로 안전한 에이전트 개발 지원
🌐 오픈 모델: 최첨단 오픈 모델, Agent Knowledge Foundation과 Kumo, CUDA-X 제공
⚙️ 산업 혁신: 의료, 통신, 제조 등 다양한 산업 분야에서 AI 적용 확대
🚗 피지컬 AI: Hyperion과 Alpamayo로 자율주행 생태계 확장하고, Isaac GR00T 기반 휴머노이드 개발 지원
synabreu retweeted
NVIDIA's Vera Rubin NVL72 could train large mixture-of-experts AI models using 75% fewer GPUs than Grace Blackwell.
Training a 10T-parameter model on 100T tokens in one month would just require roughly 16,000 Rubin GPUs, compared with 64,000 GB200 GPUs.
synabreu retweeted
Nvidia also reduced the NVLink bandwidth per GPU from 1.8 TB/s (3.6TB/s bidi) to 1.5 TB/s (3 TB/s bidi)
AMD now has a higher HBM bandwidth (23.3 TB/s) as well as a higher scale-up bandwidth (1.8 TB/s)
synabreu retweeted
Congrats to the winners of the AI Grand Prix in Singapore 🏆
Everyone had 24 hours to build, with a combination of individual challenges and a team project.
Big thanks to @Coupang and @Carousell for the partnership.
synabreu retweeted
PROFIT ALERT🚨: AMD @vllm_project can generate up to $3/hr/chip in pure profit serving the open-source Kimi K3 model, even after accounting for the rapid increase in chip rental prices post-Claude Code.
Kimi K3 2.8T is a great model for AMD, as it has over 2.8 trillion parameters that can take advantage of AMD's 288GB per chip. B200 CANNOT fit Kimi K3 in a single server, so it has to use tensor parallelism or pipeline parallelism for communication across 400Gbit/s InfiniBand/Ethernet, which is 18x slower than the BW within an NVLink domain.
synabreu retweeted
243 people signed up for a 60-person AI meetup in my hometown. I honestly didn't expect this..
After attending OpenAI DevDay in San Francisco, I wanted to bring that same energy back to Suwon, South Korea.
Yesterday, 59 of us came together. Developers, designers, creators, students, and people who had never written a line of code.
We shared what we'd built with Codex, tried live AI demos, and connected over our love of AI.
I was worried people might feel awkward meeting strangers. Instead, their biggest complaint was that we didn't have enough time to talk.
That meant more to me than anything.
AI communities don't have to exist only in major tech hubs.
Sometimes, all people need is a place to find each other.
And this is just the beginning. ❤️
Huge thanks to OpenAI, our speakers, volunteers, and everyone who made this possible!
synabreu retweeted
One bottleneck for the AI buildout is skilled labor
US manufacturers are working existing staff longer and paying up for new ones:
- Hours per worker are outpacing headcount growth in 7 out of 10 sectors
- Workers who switch jobs are getting ~10% pay increases, about double the ~5% for those who stick around
Charts of the Week: a16z.news/p/charts-of-the-we…
Since 2019, imports of AI hardware have grown faster than US production in every category
US production vs. imports:
- Computers: +46% vs. +366%
- Comms equipment: +87% vs. +175%
- Semis: +76% vs. +131%
- Grid & electrical equipment: -7% vs. +69%
Charts of the Week: a16z.news/p/charts-of-the-we…
synabreu retweeted
Finding it very educational and interesting to learn about the hardware layer of GPUs: how they are structured, the obvious advantages and disadvantages bs CPUs.
This is from Modal’s excellent “GPU Glossary” print book @can gave me (the PDF is free here: modal.com/gpu-glossary)
synabreu retweeted
vLLM 101
> Install vLLM, serve a model
> Predict how many users fit on your GPU
> Pick the one or two flags that matter for your traffic
> Decide from a measurement when speculative decoding, FP8, or a compile level pays
> Debug a slow or broken server
aiengineeringfromscratch.com…