# Ben Lerner

> Ben Lerner is the co-founder and CEO of Espresso AI, a New York startup building what it calls the first neural compute optimizer - software that uses large language models to rewrite SQL and reallocate compute so companies cut their Snowflake and Databricks bills by up to 70%. A former Google and DeepMind engineer who worked on search, storage infrastructure, and early code-LLM research, Lerner left Big Tech to merge the two halves of his career - machine learning and systems performance - into a single product. Espresso launched from stealth in May 2024 with over $11 million in seed funding from Nat Friedman, Daniel Gross, and FirstMark's Matt Turck.

- **Role:** Co-Founder and CEO at Espresso AI
- **Organizations:** Espresso AI, Google, Google DeepMind, Uber, Data Nitro, MIT
- **Nationality:** American
- **Education:** Massachusetts Institute of Technology (MIT)
- **Known for:** Co-founded and leads Espresso AI, billed as the first neural compute optimizer built to leverage LLM code analysis and generation, Raised over $11 million in seed funding from Nat Friedman, Daniel Gross, and FirstMark's Matt Turck, Built a product that reduces Snowflake compute costs by up to 70-80% and deploys in roughly 10 minutes

## Career timeline

- **early career** — Joined Data Nitro, a startup integrating Python into Excel, straight out of school
- **mid career** — Worked on distributed systems at Uber
- **~3 years at Google** — Worked on Google Search doing machine learning and natural language processing
- **later at Google** — Moved to Google's Storage Infrastructure org building high-performance code underlying Google Cloud
- **end of Google tenure** — Did research with DeepMind on reinforcement learning for efficient multi-GPU neural network training and early code-LLM models
- **2023** — Left Google to found Espresso AI, merging his ML and systems-performance backgrounds
- **2024-05** — Espresso AI emerged from stealth with over $11M in seed funding
- **2025** — Espresso AI launched its Agentic Lakehouse for Databricks, expanding beyond Snowflake

## Achievements

- Co-founded and leads Espresso AI, billed as the first neural compute optimizer built to leverage LLM code analysis and generation
- Raised over $11 million in seed funding from Nat Friedman, Daniel Gross, and FirstMark's Matt Turck
- Built a product that reduces Snowflake compute costs by up to 70-80% and deploys in roughly 10 minutes
- Assembled a founding team of ex-Google, ex-Apple, and ex-MIT engineers
- At Google, contributed to Search ML/NLP and the storage infrastructure underlying Google Cloud
- Did DeepMind research on RL for efficient multi-GPU training and early code-LLM models
- Served as a judge for the AI Poverty Challenge

## Latest updates

- **2024-05** — Espresso AI emerged from stealth with $11M+ seed funding to tackle the cloud cost crisis, starting with Snowflake SQL optimization.
- **2025** — Espresso AI launched an Agentic Lakehouse for Databricks and was described as building 'Kubernetes for Snowflake.'

## Links

- Website: https://espresso.ai
- LinkedIn: https://www.linkedin.com/in/espressoben/
- Twitter/X: https://x.com/ben_lern
- YouTube: https://www.youtube.com/watch?v=LPCgyQ7gBiM

---

Profile page: https://yespress.io/ben-lerner
Published by YesPress — https://yespress.io
Last updated: 2026-06-25