Background
The visible trail is a technical-programme-management one on data rather than a research one: recruiting for a Senior TPM Data role, sourcing expert AI tutors, and posting about Gemini. That pattern - job reqs, tutor sourcing, mid-training and RL data - is the profile of someone running data supply for a post-training org rather than someone designing models. Nothing else about her career or education is publicly verifiable.
What they run now
Inferred: sourcing and running the human-data supply that feeds mid-training and reinforcement learning - expert tutor panels, task design, throughput and quality against a post-training roadmap. The 'Expert AI Tutor' request-for-services post is the most concrete signal: it is the language of buying expert human labour, not of labelling at volume.
What they buy
Inference. Mid-training and RL data: expert-written demonstrations, preference pairs, RL environments and graded rollouts, plus the expert panels to produce them. The 'request for services - Expert AI Tutor' framing suggests she contracts specialist individuals or supplier panels directly rather than buying generic annotation volume.
Career
- presentData (mid-training / RL), Employer not confirmed (Google DeepMind indicated by 2023-2024 posts) — Headline only; org unverified.
