Denis Savenkov

Member of Technical Staff at @perceptroninc

[Denis Savenkov @DenXX]

The interesting part isn’t just open-vocabulary detection. It’s treating perception as an iterative process—looking closer when needed, building evidence, and adapting to the task instead of relying on a fixed detector.

[Perceptron AI @perceptroninc]

Today we're releasing Perceptron Agentic Detection: localize anything you can describe in natural language or show examples of.

[Denis Savenkov @DenXX]

Really excited the share one new Perceptron Mk1 model, significantly improved perception and now with video understanding.

[Perceptron AI @perceptroninc]

Today we're releasing Perceptron Mk1: frontier video and embodied reasoning.

[Denis Savenkov @DenXX]

You can now integrate better vision capabilities into your agentic workflows using our Perceptron MCP

[Perceptron AI @perceptroninc]

Despite having vision, most AI agents still struggle to see. General-purpose multimodal models are powerful, but they’re expensive for every visual task. We built something better: Perceptron's MCP gives any agent stronger vision capabilities through Isaac with far lower cost.

[Denis Savenkov @DenXX]

Isaac 0.2 providing vision to OpenClaw

[Perceptron AI @perceptroninc]

We plugged our vision agents (Isaac 0.2) into OpenClaw, turning vision into action.

[Perceptron AI @perceptroninc]

Data curation methods collapse quality into a single score, but quality is multidimensional. We are excited to share SkillRater — a multimodal extension to DataRater. By decomposing filtering into capability-aligned raters, we outperform monolithic scoring across all dimensions with near-orthogonal learned signals

[Perceptron AI @perceptroninc]

Today we’re open-sourcing a preview of our two new models in the Isaac family: hybrid-reasoning 2B and 1B-parameter best-in-class vision-language models.

Weights → https://t.co/1WgHMDfCST
Blog → https://t.co/8MOLPKpUhO
Demo → https://t.co/sAKt5dnZ6U

[Perceptron AI @perceptroninc]

👀

[Perceptron AI @perceptroninc]

Perceptron’s platform is here — built for Physical AI

Developers can now use Isaac-0.1 or Qwen3VL 235B via:

Perceptron API — fast, reliable multimodal intelligence
Python SDK — simple, grounded prompting for vision + language

Build apps that see and understand the world.

[Perceptron AI @perceptroninc]

Today we’re releasing the technical details behind Isaac 0.1. We built Isaac to demonstrate that with the right principles, a simple recipe can reach competitive performance with a small model.

Check out the report at: https://t.co/Y1Ys2NHswX

[Perceptron AI @perceptroninc]

Designing multimodal systems is challenging. To build with the community we’re sharing the design behind TensorStream - a convenient tensor-like interface for interleaved multimodal data. Internally our training and inference code is built on top of this. https://t.co/qnkePqXhy6

[Denis Savenkov @DenXX]

@perceptroninc This screenshot shows the detection bounding box a bit better.

[Denis Savenkov @DenXX]

Super excited to share what we've been up to at @perceptroninc Today we've released Isaac 0.1 — our first perceptive-language model: 2B params, but highly capable. Check it out: https://t.co/ohlQOUwxm3

[Perceptron AI @perceptroninc]

1/ Introducing Isaac 0.1 — our first perceptive-language model. 2B params, open weights. Matches or beats models significantly larger on core perception. We are pushing the efficient frontier for physical AI.
https://t.co/dJ1Wjh2ARK

[Denis Savenkov @DenXX]

Highly recommend the https://t.co/bdkVi6pbc3 weekly summary to keep up with the field of ML. They're using AI to identify the top publications and breakthroughs, it's a game-changer.

[Eugene Agichtein @eugeneAgichtein]

Honored that our 12-year-old work at Microsoft on incorporating user behavior signals into ranking (with Susan Dumais and Eric Brill) was recognized with SIGIR ToT award. Also, happy to see that 12 years later there is still lots of room for improvement :-) https://t.co/yAjHxcSiUm

[Milad Shokouhi @invertedindex]

Ahmed Hassan Awadallah and I have one open research internship position this summer. If you're interested, please let us know and kindly share your CV!

[Simon Brunning @brunns]

Finally managed to quit vim.

[Fernando Diaz @841io]

I am humbled to receive an award named in honor of Karen Spärck Jones, whose contributions continue to be relevant to information retrieval researchers and engineers. https://t.co/SSgCKhZLux

[Denis Savenkov @DenXX]

@eugeneAgichtein I'm really grateful to have such an advisor and mentor! Also thanks to the committee: Prof. Jinho Choi, Prof. Li Xiong and @scottyih

[EmoryCSI @emory_csi]

Congrats to Dr. Denis Savenkov @DenXX on defending PhD on Question Answering w User Generated Content! He will join FB as research scientist