# Mahesh Sathiamoorthy

> Mahesh Sathiamoorthy is the co-founder and CEO of Bespoke Labs, an applied AI research company building reinforcement-learning environments, data-curation systems and evaluation tools for more reliable AI agents. A former staff software engineer at Google DeepMind, he moved from distributed storage research at USC to large-scale recommendation systems at YouTube, then left Google in 2023 and started Bespoke Labs in 2024 with Alex Dimakis.

- **Role:** Co-Founder and CEO at Bespoke Labs
- **Organizations:** Bespoke Labs, Google DeepMind, Google Brain, YouTube, University of Southern California, Indian Institute of Technology Kharagpur
- **Education:** B.Tech (Honours), Indian Institute of Technology Kharagpur, M.S. and Ph.D., University of Southern California
- **Known for:** Co-founded Bespoke Labs and serves as CEO., Says he was the first to productionize TPUs for recommender systems at YouTube., Co-authored Recommender Systems with Generative Retrieval, published at NeurIPS 2023.

## Career timeline

- **2008-2013** — Graduate study at the University of Southern California, including research on coding for distributed storage and dynamic networks.
- **2013** — Co-authored XORing Elephants, a VLDB paper on locally repairable codes for Hadoop storage.
- **2014-2023** — Worked at Google, later as a staff software engineer at Google DeepMind, across Google Brain and YouTube recommendation systems.
- **2019** — Co-authored Recommending What Video to Watch Next, describing a large-scale multitask ranking system for YouTube.
- **2023** — Co-authored work on generative retrieval and semantic IDs for recommender systems, and left Google DeepMind to start something new.
- **2024** — Started Bespoke Labs with Alex Dimakis, focused on data curation and reinforcement-learning environment curation.
- **2025** — Bespoke released Curator and helped originate OpenThoughts; Sathiamoorthy appeared on TWIML to discuss reinforcement learning for custom agents.
- **2026** — Bespoke Labs announced $40 million raised across seed and Series A rounds to expand its work on reliable AI agents.

## Achievements

- Co-founded Bespoke Labs and serves as CEO.
- Says he was the first to productionize TPUs for recommender systems at YouTube.
- Co-authored Recommender Systems with Generative Retrieval, published at NeurIPS 2023.
- Co-authored the KDD 2023 Data Science Track best-paper winner on training stability for multitask ranking models.
- Co-authored XORing Elephants, a 2013 VLDB paper on locally repairable codes for big-data storage.
- Helped initiate OpenThoughts and led a company that became a core contributor to Terminal-Bench.
- Reports receiving more than 50 Google peer and spot bonuses, more than five performance awards, a best-paper award and a Feats of Engineering award.

## Latest updates

- **2026-07** — Bespoke Labs announced $40 million raised across seed and Series A rounds, led at Series A by Wing Venture Capital, with plans to deepen its work on reinforcement-learning environments and data curation.
- **2026-07** — Sathiamoorthy said Bespoke would continue building toward a data research lab focused on post-training and reliable agents.
- **2026-05** — Participated in a BlackRock SF Applied AI Forum panel on architecting and orchestrating agentic AI.
- **2026-04** — Participated in a Snowflake panel on post-training agents and sandbox environments.
- **2026-03** — Gave a talk titled Pushing the Frontier Forward at UC Berkeley.

## Links

- Website: https://smahesh.com/
- LinkedIn: https://www.linkedin.com/in/smaheswaran
- Twitter/X: https://x.com/madiator

---

Profile page: https://yespress.io/mahesh-sathiamoorthy
Published by YesPress — https://yespress.io
Last updated: 2026-08-18