Brooklyn, New York, United States
I joined Stack Overflow just a few weeks before ChatGPT launched. What followed was the most compressed product education of my career — navigating platform disruption in real time, building AI features that had to meet developer-level quality standards, and ultimately turning a looming threat into a new business line. My team built Stack Overflow conversational search through four iterations — from a chat UI on keyword search, to semantic search, to GPT-4 fallback, to a full RAG implementation with attribution. We developed evaluation frameworks using subject-matter experts, gold-standard datasets, and production signals to separately measure accuracy, relevance, and completeness. When the product couldn't clear the quality bar developers expected, we made the hard call to roll it back. That wasn't a failure — it was the foundation for what came next. I then led the creation of Stack Overflow's data licensing business, partnering with OpenAI, Google, and Microsoft to integrate 14M+ community Q&A pairs into their AI products. My team built industry benchmarks that proved Stack Overflow data measurably improved LLM accuracy on technical questions — work that helped grow the business from idea to mid-eight figures in ARR in under 15 months. Before Stack, I was Head of Product at Chegg Skills and the first product manager at Thinkful, where I helped guide the company through an $80M acquisition. I also teach Defining Outcomes for Product Talk and advise early-stage AI companies on product strategy. Today I work with product leaders navigating the same shift I lived through at Stack Overflow. I help teams build eval strategies, define AI product roadmaps, and ship features that retain users — not just demo well. If your AI features spike at launch and flatline by day 30, let's talk.
I help product leaders level up as they transition into new roles — building the skills that matter most in the first year: setting outcomes, leading teams, managing stakeholders, and making tough prioritization decisions. Increasingly, that work includes helping leaders navigate how AI is reshaping the products and teams they're responsible for.
I teach the Defining Outcomes course, which helps product teams translate business goals into measurable product outcomes. I also support community members in applying Continuous Discovery methods and evaluating generative AI implementations. Product Talk helps digital product teams gain valuable insights from their customer interviews, run effective product experiments, and drive product outcomes that create value for their customers and their businesses.
Salley is a research lab and applied AI company advancing human potential in the age of AI. I advise the executive team on product strategy, AI implementation, go-to-market, and fundraising.
Led the cross-functional team that built Stack Overflow's data licensing business — the product line that turned the company's 14M+ Q&A corpus into a revenue engine for the AI era. Partnered with OpenAI, Google, and Microsoft to integrate community knowledge into their AI products and models. Built technical benchmarks with subject matter experts that proved Stack Overflow data measurably improved LLM accuracy on technical Q&A — with controls for data leakage and independent validation. Designed partnership structures that created virtuous cycles: labs improved their products with Stack's data while driving engagement back to the developer community. Grew the business from concept to mid-eight figures in ARR in under 15 months.
Led product strategy across Stack Overflow's full portfolio (B2B, B2C, B2B2C) during a period of rapid strategic repositioning. Identified the data licensing opportunity that became the company's fastest-growing business line. Defined strategic positioning and new monetization paths as developer workflows shifted toward AI tools.
Stood up and led the Overflow AI team — a cross-functional group tasked with exploring what was newly possible for developers with generative AI. The team ran 50+ experiments in three months, shipping prototypes weekly and sharing demos across the org. Led the development of conversational search through four major iterations: keyword search → semantic search → GPT-4 fallback → RAG with attribution. Built eval infrastructure from scratch — subject matter expert panels, golden data sets of 250+ questions, and automated quality checks scoring accuracy, relevance, and completeness independently. Made the decision to sunset conversational search when it couldn't meet the quality bar developers expected. The eval frameworks and technical learnings from this work directly powered the data licensing benchmarks that followed.