Greater Seattle Area
I build AI teams and ship AI products. At Microsoft, I lead AI/ML for Real-Time Intelligence in Fabric. Shipped Operations Agent and real-time anomaly detection in my first six months. Before that, I led research and engineering for Google Gemini during its launch year, creating the evaluation frameworks that drove 10x quality improvement while cutting eval costs by 30%. My through-line is AI evaluation. I've built eval systems for LLMs, recommendation engines, and now autonomous agents. I'm currently writing a book on the topic. Previously: Twitter (Home Timeline personalization), Microsoft (Azure, Windows), Check Point (security). Started as an engineer, still think like one.
Leading AI/ML strategy and execution for Real-Time Intelligence within Microsoft Fabric. Shipped Operations Agent and real-time anomaly detection in the first 6 months. Built and grew the team from 0 to 10 to make it happen. Designing evaluation frameworks for autonomous agents, extending LLM eval methodology into agentic systems.
Advising an early-stage startup rethinking how recruiting works. Shaping ML strategy, model architecture, and build-vs-buy decisions for AI infrastructure.
Led research and engineering for Gemini during its launch year, owning LLM evaluation methodology and post-training data strategy (SFT, RLHF). Built evaluation frameworks that drove a 10x improvement in Gemini's overall quality while reducing eval costs by 30%. Developed novel evaluation approaches for human and LLM evals that became standard practice across Gemini. Operated at the intersection of DeepMind research and Google product, translating research advances into shipped model improvements.
Led the ML and data science function for Google Workspace's SMB growth. Built production ML models for churn prediction, upgrade propensity, and customer acquisition that drove targeting and intervention decisions across the business. Designed real-time anomaly detection systems monitoring Workspace's core metrics. Owned experimentation strategy for major product launches.
The Design Thinking Program at the Tombolo Institute at Bellevue College is intended to train you and develop your abilities to apply innovative Design Thinking principles to increase ROI at your organization.
Led three data science teams driving model improvements, experimentation, and evaluation for Twitter's Home Timeline and personalization. What hundreds of millions of users saw when they opened the app. Drove the initiative to redefine Twitter's top-line business metrics.