Profile banner
Ilya Kulyatin

Ilya Kulyatin

AI entrepreneur and product builder; founder of Foundry Labs

ikulyatin
Tokyo, Japan
Joined August 2025
Favicon imagefoundrylabs.co.jpFavicon imagetokyoai.jpFavicon imagemedium.comFavicon imagegithub.comFavicon imagex.comFavicon imageventurecafetokyo.org

Network

12.3K connections

Summary

Founder and product-focused AI engineer: Ilya founded and leads Foundry Labs, positioning the firm as a senior enterprise AI engineering studio that builds production-grade systems—custom LLMs, robotics/edge AI, hardware acceleration—and delivers end-to-end product solutions for regulated and enterprise customers. co
Community builder and events organiser in Tokyo: he runs and curates the Tokyo AI (TAI) community—organising meetups, AMA series, hackathons and invite-only TAIT sessions—to grow Japan's AI ecosystem and partner with industry for practical sessions on deploying LLMs, robotics and applied ML. tokyoai+1
Quantitative researcher and crypto risk specialist: co‑founder and Head of Research at Cloudwall Capital, he has a background in quantitative research and fintech, building institutional crypto risk tooling (Serenity) and previously held quantitative roles across market-making and asset management. blockworks+1
Public communicator and technical writer: maintains an active Medium publication with technical posts on factor investing, market-impact modelling and crypto risk, and uses public developer and social profiles to share research, event organisation and technical commentary. medium+3

Work

Education

Projects

Writing

Market impact modeling for Bitcoin

December 1, 2025

Technical Medium essay describing market impact decomposition (temporary vs permanent), the I‑Star model application to BTC, data requirements and empirical modeling using CEX order-book data.

Favicon imagemedium.com

A primer on factor investing and frameworks to identify and extract risk premia

May 1, 2022

Medium article explaining factor-investing frameworks, identification and extraction of risk premia, and applying these ideas to crypto-specific factors.

Favicon imagemedium.com

Model-based Deep Reinforcement Learning for Dynamic Portfolio Optimization

January 1, 2019

Co-authored academic paper presenting a model-based deep RL architecture for dynamic portfolio optimization, including an infused prediction module, data augmentation module and behavior cloning component; evaluated on historical financial data.

Favicon imagearxiv.org

Hobbies

Community building and event organization for AI (meetups, AMAs, hackathons). tokyoai+1

Notes

FJFederico Jarachsaid
IKIlya Kulyatin
is someone they turn to for insight on whether an AI offering actually fits real lab workflows.

Jan 26, 2026