Weiwei Sun

Working on 3D for Amazon Just Walk Out (JWO). Applied Scientist @Amazon, Ph.D.@UBC, intern @Amazon, Intern @NvidiaAI, Intern @Google. Opinions are my own.

[Jun Gao @JunGao33210520]

1/5 🚀 Thrilled to open-source OSCAR 🤖 — an action-conditioned world model for robotics, led by the visiting student in my group @wuzy2115! It generalizes across different robot embodiments with precise action controllability. All trained on a single GH200 GPU, and outperforms existing open-sourced baselines, which have larger model capacity and need more compute.

Everything is public, including training data.
📄 Paper: https://t.co/0KokZcPhP5
🌐 Project: https://t.co/g3SGfAHI76
💻 Code: https://t.co/Hv81Yjy5kV
🤗 Robot data: https://t.co/p6g967kbXg
🤗 Human data: https://t.co/5f5pMnZGrc
🤗 Weights: https://t.co/HRAloQeh8t

#Robotics #WorldModels #AI #OpenSource

[Weiwei Sun @wsunid]

Hey, folks who just finished CVPR, 3D is not dead yet at least in Amazon JWO (Autonomous store) where we still care 3D reconstruction and camera stuff. Please drop me an Email if you are looking for a FTE since we are looking for a scientist for 3D (SfM, GFM) now!

[Weiwei Sun @wsunid]

Time of getting a work permit in Canada is ~259days (>8mons). I feel sorry for PhD students who need a work visa for summer internship. I don’t think tech companies can wait for 8mons.

BTW, JWO is recruiting interns (US or Canada?) in 3D/VLM/VDM. Shot me your CV.

[Weiwei Sun @wsunid]

TTT seems increasingly popular!

[Haian Jin @Haian_Jin]

Spatial reconstruction is a long-context problem: real scenes come with hundreds of images. But O(N²) transformer-based models don’t scale efficiently.

Introducing: 🤐ZipMap (CVPR ’26): Linear-Time, Stateful 3D Reconstruction via Test-Time Training (TTT).

ZipMap “zips” a large image collection into an implicit TTT scene state in a single linear-time operation. The state will then be decoded into spatial outputs, and can be queried efficiently for novel-view geometry and appearance (~100 FPS)

ZipMap is not only much faster (>20× faster than VGGT), but also matches or surpasses the accuracy of all SOTA models.

[Weiwei Sun @wsunid]

@maxjiang93 Very nice and solid work! Congrats Max!

[Weiwei Sun @wsunid]

@GoogleDeepMind I’m curious what’s the plan for code releasing? We can’t know much till we try the model.

[Weiwei Sun @wsunid]

@TransLink @tpoole00 Good job Vancouver SkyTrain🙂

[Weiwei Sun @wsunid]

@anshulkundaje Citation count is not a perfect metric but it’s still a thing.

[Weiwei Sun @wsunid]

Hi, I will be at @NeurIPS25_San_Diego Please reach out and always happy to chat about 3D vision, simulation, generative modelling etc and their challenges in real-world applications!

[Weiwei Sun @wsunid]

Yep!

[Kosta Derpanis (sabbatical in Zurich) @CSProfKGD]

How does @kwangmoo_yi only have 3.8k followers?!?! Folks, Kwang keeps surfacing absolute gems. Drop him a follow.

[Weiwei Sun @wsunid]

@CVPR Thanks! 😊

[Weiwei Sun @wsunid]

Hi @CVPR , is there any way to check if all authors of my submission have the complete profile? Is it mandatory to get it done by today? Thanks!

[Weiwei Sun @wsunid]

@ducha_aiki Interesting—similar to what I found with more recent Pi3 and mapanything. I’m now curious if we able or not to get rid of RANSAC with data and model scaled up infinitely.

[Weiwei Sun @wsunid]

@tedlasai Cool work! While it has the better capability than RGB sensors, do you know if these multi-spectral sensors are scalable? Price, lifespan, energy consumption etc.

[Weiwei Sun @wsunid]

@mashtaktash Congrats Maria!

[Weiwei Sun @wsunid]

@zhoubolei @UCLA @UCLAengineering Congrats Bolei! And I like your dog!

[Weiwei Sun @wsunid]

@yongyuanxi Agree! Geomatics is super interesting and practical. Though, a lot of my peers (yes, I also hold BSC/MSC degrees in this area) changed their focus to science problems such as climate change and biodiversity. Geomatics turns out to be the tech/tools behind their research.

[Weiwei Sun @wsunid]

Just walk out(JWO) — a very cool tech that enables cashier-less store! Just go to the session and learn more!

[Amazon Science @AmazonScience]

Thanks for checking out our live demos at #CVPR2025 today! Here are the others we’re hosting this weekend:​​​​​​​​​​​​​​​​

Saturday, June 14:
◽️10:30-11am: Multi-Modal AI - Just Walk Out & Visual Reasoning
◽️1:30-2pm: Audio/Video Generative Intelligence at Prime Video
◽️2-2:30pm: Vulcan Stow/Pick

Sunday, June 15:
◽️10:30-11am: Multi-Modal AI - Just Walk Out & Visual Reasoning
◽️11-11:30am: Audio/Video Generative Intelligence at Prime Video

[Weiwei Sun @wsunid]

Kudos to Colin (my first mentee)🎉🎉🎉 Learn more about surface reconstruction from large-scale point cloud? Check it out below👇👇👇

[Andrea Tagliasacchi @CVPR @taiyasaki]

📢📢📢 NoKSR: Kernel-Free Neural Surface Reconstruction via Point Cloud Serialization

https://t.co/1rxX7UDc0q
https://t.co/iuA7HJU49E
#3DV2025

TL;DR: neural reconstruction with a simpler architecture (no linear systems to solve), and up to 3x speedup vs. voxel-based methods!

[Weiwei Sun @wsunid]

@yongyuanxi Indeed, I spent most of last two months in figuring out bundle adjustment under weak constraints.

[Weiwei Sun @wsunid]

@yongyuanxi +1