@williamwei profile picture
William Wei @williamwei

AI First Technologist, former Foxconn & MIH CTO / Apple & NeXT engineer

Joined June 2007
  • Tweets 288
  • Following 304
  • Followers 382
26 Photos and videos
Pinned Tweet
William Wei@williamwei
15 Mar 2025
Edge AI needs acceleration medium.com/p/building-ai-rea…
129
So, it turns out #harness engineering needs to match with the behind the scenes #model, harness designed for the frontier model falls apart when coupled with a small to medium model. It seems harness is not just a wrapper, it could be the moat.
31
#HTX301 and #Skymizer Google them
1
69
Beyond TurboQuant: Why Software–Hardware Co-Design Defines the Next Era of AI Inference medium.com/p/beyond-turboqua…
26
shell-WG3V
16
The bottom logic of Meta's hashtag #Moltbook acquisition is about hashtag #Agent hashtag #Identity and the linking network of human identity. linkedin.com/posts/williamwe…

#moltbook #agent #identity | William Wei

The bottom logic of Meta's #Moltbook acquisition is about #Agent #Identity and the linking network of human identity.

linkedin.com
41
#Agentic #AI update: #Automotive digital - youtu.be/HpZJ-xpLZgU physical - youtu.be/0iDn8kUE9rg #Smarthome #AIoT digital - youtu.be/MO85tOecHFM physical - youtube.com/shorts/nUeOmaQ3c… #Humanoid digital - youtu.be/oLgpTWXuqD8 youtu.be/-b7n1YfyGEI

Vehicle Simulator with vehicle voice assistant (KITTWeb)

Vehicle Simulator with vehicle voice assistant (KITTWeb)1. Vehicle...

youtube.com
59
The only solution/option left for AI #safety is #AI!
34
In short, Coding Agent is Context Engineering.
57
So vibe coding turned into Context Engineering, which makes much more sense. The LLMs's limitations are the pre-training, context window and all the external tools, so what you will be responsible for, highly depends on what your provide and arrange in context window? @karpathy
1
2
114
Automotive #LLM is the software 3.0 version of the car OS !
49
William Wei retweeted
Skymizer Taiwan Inc.@skymizer
27 May 2025
COMPUTEX 2025 may be over, but the conversations continue! Check out this interview with our CMO @williamwei , featured by DIGITIMES 🎥👇 🔗 youtube.com/watch?v=JG74ulVu… #Skymizer #COMPUTEX2025 #LLM #AIchip #HyperThought #EdgeThought #DIGITIMES

Straight from COMPUTEX 2025 – Highlights from MSI, Infineon, Skymizer...

Welcome to DIGITIMES’ special coverage of COMPUTEX 2025! This year’...

youtube.com
1
1
251
Is #MCP really needed? Do we need to invent another kind of STOP sign or Traffic Lights just for Autopilot? Is MCP a temporary thing because current #LLMs are not smart enough?
57
Why do we want #AGI (being sentient or conscious) to compete with the current human population? What we want is highly specialized & safe artificial super intelligence (#ASI) as tools for human population. Agree?
101
The future is Agentic AI! And, agents are replacing Apps… #Agentic AI in #Automotive
0:29
1
4
427
Traditional software engineering write code to solve problems, but AI uses end-to-end training to solve complex problems, especially those non-deterministic hard problems.
1
66
And, it turns out, the non-deterministic domain is much larger than the deterministic domain, in the real world we live in. By training AI models to generate codes is to mimic the traditional software engineering using end-to-end approach. makes sense?
49
Back to the root of Skymizer in open source and hacking community
Skymizer Taiwan Inc.@skymizer
1 Aug 2024
Skymizer is proud to be a sponsor of COSCUP this year! Through our sponsorship, we aim to provide more resources to the open-source community and work together to create a better software ecosystem. Don't miss out on the 2024 COSCUP happening this weekend!coscup.org/2024/zh-TW/sessio…
105
meet-global.bnext.com.tw/art…

Skymizer Unveils Revolutionary EdgeThought Accelerator for On-Device LLM Inferencing｜Meet Global

Skymizer has just introduced EdgeThought, a groundbreaking AI accelerator tailored for on-device LLM inferencing. This innovative solution, powered by advanced compiler technology, is set to redefine...

meet-global.bnext.com.tw
50
By eliminating the need for training software and hardware, #EdgeThought emerges as the premier choice for #on_device #LLM #inferencing. as the "game-changer for on-device LLM inferencing," compatibility with mature silicon class and specialty memory components,
45
Load more