Aditya Mehta

cs @Caltech

[Aditya Mehta @amehta_05]

@anvisha using voyager + astra to make 3b1b style videos to explain core concepts / proofs for my cryptography course:

the results are insanely good. crazy to think that this is the worst itll ever be

[Anvisha @anvisha]

Introducing Voyager: the open harness for S-tier video, graphics and games.

The Codex for creative work.

Most AI tools are built for coding. Voyager is designed to get the best creative results from models like Opus, Astra and DeepSeek.

It comes with free graphics, music and video editing tools - or, it can drive Blender, Resolve, After Effects, Ableton and 100+ others.

Everything in this thread was made with Voyager. A few of our favorites:

[Aditya Mehta @amehta_05]

Experts in multimodal MoEs specialise by domain / modality!

Check out the work below led by @marsilidamiano and @RaphiKang to see how we match experts directly from weights (no dataset required!); and update only them to match fft / lora while being 4x / 3x faster (!)

[Damiano Marsili @marsilidamiano]

What internal structure do experts in multimodal MoEs develop during pre-training? Can we put this structure to practical use to make model adaptation significantly faster?
In our new blog post and paper with @RaphiKang and @georgiagkioxari we discuss this and more 👇

[Anvisha @anvisha]

Launching today: your design partner in Slack.

At mention Moda in any thread. It reads the full conversation and builds client-ready materials on the spot.

Sales decks. Reports. Social posts. Even memes.

Try it now: https://t.co/rT4fFjuU3u

[Aditya Mehta @amehta_05]

Excited to launch my first project from my internship @trymoda!

[Anvisha @anvisha]

Launching today: Layerize any image

Drag in an image and Moda will rebuild layers that you can edit directly.

Early users have been saying text reproduction is way better than Canva or Photoshop.

See it in action:

[Anvisha @anvisha]

Introducing the Moda API: the design layer for agents, apps, and workflows.

Embed presentations, reports, and design creation right into your product or agent.

See what people are building with it 👇

[Aditya Mehta @amehta_05]

@anvisha Excited for the launch!

[Anvisha @anvisha]

We raised $7.5M to kill AI slop.

Introducing Moda: the world's first design agent with taste.

RT+ comment “Moda” and we’ll design your brand for FREE.

[Damiano Marsili @marsilidamiano]

Happy to announce TWIN was accepted to #CVPR2026!

We introduce a large-scale VQA dataset for fine-grained understanding in VLMs and show that post-training on this data improves fine-grained visual reasoning for OOD domains!

See everyone in Denver! 👋🏔️

[Damiano Marsili @marsilidamiano]

(1/6) Do these images show the same vacuum cleaner? They are certainly similar, but a human will notice the differences in dustbin geometry, design, and color accents. In contrast, open-source VLMs struggle at this task. Our recent work TWIN poses the question: Can we fix this?

[Damiano Marsili @marsilidamiano]

(6/6) This work was done in collaboration with two amazing undergrads @Amehta_2005, @rlin232, and my advisor @georgiagkioxari !
Project page: https://t.co/gqkqswlK4U
Paper: https://t.co/MgI17D14YA