Human-data operator who led human data operations at OpenAI, then became a Managing Director at Mercor. His LinkedIn now shows Google DeepMind, i.e. he appears to have moved back to the buyer side.
medium confidenceAt Mercor since 2024Updated 2026-09-19
Background
Software engineer by training (BS Software Engineering, Iowa State). Worked at Microsoft, Amazon, Boeing, MathWorks, Twilio and Engagio, then at Scale AI (field engineering manager → Head of Field Engineering → Director of Computer Vision Engagements & Field Engineering), then OpenAI (per SF Standard, head of human data operations), then Mercor.
What they run now
If he is at Google DeepMind, he is likely again on the buying side of human data (speculation — role not confirmed).
Career
presentUnknown, Google DeepMind — per LinkedIn headline in search results; not independently confirmed
—Managing Director, Mercor
—Head of human data operations, OpenAI
—Director of Computer Vision Engagements & Field Engineering, Scale AI
—Iowa State University, BS Software Engineering
Recent posts
4 posts archived · most engaged first, then the latest
For those who don't know, my guilty pleasure is playing video games building factories, towns, and transportation networks. (Factorio, Anno 1800, Cities Skylines, for example)
Getting multi-modal LLMs to be able to play these games seems like a potential holy grail of computer-using agents, long-horizon reasoning, context management, and a lot more.
I already started tinkering with automating Factorio and got somewhere (there was a cool paper on that one already providing the game engine, the agents write code to interact with the environment) but if anyone else knows of companies or ideas trying to do this, I'd love to learn more.
From a fun idea I had post-OpenAI to the New York Times, REAL Evals has come a long way!
I'm thankful to Div Garg for partnering with me on it as one of the first believers, and for the amazing work done by the engineers I had hired to bring their front-end expertise to bringing my vision to life, Federico Gabriel López, Michael Lara, Tomás Abraham
I really wanted this vision of practical evals to exist in the world, and I think it's fair to say, it has indeed been manifested!
lnkd.in/gjGZjgVE
I've been feeling inspired by morning meetups and amazing dinner conversations, so I thought I'd finally start hosting my own!
If you want to join me in having an amazing start to your day (or just put your email down for future breakfast / dinner / shennanigans planning) - please sign up below!
Our first breakfast will be on Wednesday, November 19
lnkd.in/dbspbp-y
That feeling when you're about to run out of your $200/mo. Cursor Ultra credits halfway through the month... And this is only using a single model, not even the fancy stuff!
Get excited, world, lots of cool things are in the works!
Potentially high: someone who ran human data at OpenAI, then sold it at Mercor, and may now sit at a buyer (DeepMind) knows exactly how labs pick and judge vendors — e.g. for multimodal/visual taste data. Verify his current role first.
How to reach
Has a public LinkedIn article reflecting on Mercor — referencing it is a natural opener. Ask how labs evaluate a small specialist vendor versus Mercor/Scale (speculation).
What we could not establish
Confirm his Google DeepMind role and when he left Mercor (likely 2026).
Miju Labs