micro1's senior AI/research leader: a Stanford PhD (Andrew Ng, Dan Jurafsky) and serial founder who gives the company research credibility with labs. Inc. describes him as a former Stanford CS professor and ex-Apple.
medium confidenceUpdated 2026-09-19
Background
Stanford PhD in CS (2015) on speech recognition and neural networks under Andrew Ng and Dan Jurafsky. Co-founded Roam Analytics (healthcare NLP, acquired by Parexel); research scientist at Semantic Machines (acquired by Microsoft); senior roles in Apple's Special Projects Group; founded and led Pointable (enterprise LLM retrieval) until its 2025 acquisition — the FORTH bio names Apple as the acquirer.
What they run now
Data-centric ML systems at micro1 — likely synthetic data generation, evaluation tooling and quality control (the 80-90% margin dataset business). Specific remit not published.
Career
presentVP of AI, micro1
2025Founder & CEO, Pointable — acquired
—Senior roles, Apple (Special Projects Group)
—Research scientist, Semantic Machines
—Co-founder, Roam Analytics — acquired by Parexel
2015Stanford University, PhD Computer Science — advisers Andrew Ng and Dan Jurafsky
On the record
Building AI companies · 2026-05
Argues LLMs changed product feasibility and development costs, so AI startups must navigate a moving goal. source
Recent posts
2 posts archived · most engaged first, then the latest
VP of AI @micro1 Trees+Tech @DeepForest.org Instructor @Stanford
•
Robotics applications are expanding fast! 🚀🤖 The micro1 robotics team just released a new blog post on how we’re approaching data annotation to make such diverse real-world applications viable. Our annotation engine searches over possible configurations to build a multi-step pipeline for each project; composing VLMs, foundation models, and human domain experts to produce the specialized, high-quality datasets new robotics applications need to achieve production.
The confluence of capable hardware, VLMs / VLAs, and reusable perception + control stacks mean you can now prototype an embodied application in the time it used to take to build a first demo. The limiting factor for specializing tasks/applications is increasingly the data required for training and evaluation. Full post with some examples below. #DataCentric#Robotics#DeepLearning#PhysicalAI#VLM
VP of AI @micro1 Trees+Tech @DeepForest.org Instructor @Stanford
•
Hunter Hayden has years serving the intelligence + defense community AND exciting projects underway for new AI systems to re-think work at places like NSA, DoD, and NSF. Join us in building the next generation of data-centric AI!
Accounts whose posts Andrew has reposted recently: Mark Esposito, PhD, micro1.
Why it matters here
The person at micro1 most likely to understand how to turn subjective judgement into reusable datasets and metrics — the technical crux of taste data. Moderate relevance.
How to reach
Academic register works: a short technical note on measuring agreement in aesthetic judgements would suit him (speculation).
What we could not establish
Start date at micro1.
X handle @AndrewLeeMaas appears in search but no evidence tied it to him — left null.
Inc. says 'former Stanford CS professor'; exact appointment not verified.
Miju Labs