Mountain View, CA, United States of America

Ilya Eckstein

This inventor holds 1 USPTO granted patent. Top assignee: Google Inc.. Active years: 2026.

IDiyas Innovation Intelligence. (2026). Inventor Profile: Ilya Eckstein. Retrieved from https://idiyas.com/inventor/ilya-eckstein-ugyaufpy

Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile


Average Co-Inventor Count = 1.0

ph-index = 1


Company Filing History:


Years Active: 2026

Loading Chart...
1 patent (USPTO):Explore Patents

Title: Ilya Eckstein: Inventor in Transformer Systems

Introduction

Ilya Eckstein is an inventor based in Mountain View, California. His patent work concerns transformer systems and methods for performing deep-learning tasks on input sequences.

Latest Patents

The provided information lists one patent associated with Eckstein: “Mixing Tokens with Spectral Transform.” The invention describes transformer systems that process an input sequence to generate an encoded output. One aspect features an encoder architecture block with a spectral transform mixing layer, which receives input-token embeddings and produces a spectral-transform output along the sequence dimension. A feed-forward layer then uses information based on the embeddings and spectral-transform output to generate an output for a subsequent processing block. The patent also covers computer programs encoded on a computer storage medium for carrying out these tasks.

Career Highlights

Eckstein works at Google Inc. His listed patent reflects work on transformer architectures and techniques for mixing token information through spectral transforms. The provided data attributes 1 patent to him.

Collaborations

The listed coworkers associated with Eckstein are James Patrick Lee-Thorp and Joshua Timothy Ainslie.

Conclusion

Ilya Eckstein’s patent work at Google Inc. focuses on applying spectral-transform token mixing within transformer systems for deep-learning tasks.

Profile summary based on public USPTO records.
Data Sources: USPTO Patent Grant XML, Patent Center, EPO & CIPO • Normalized by IDiyas Innovation Graph. Methodology & provenance architecturePlease report any incorrect information to [email protected]
Loading…