Research Operations & AI
Scaling the research function through programs, processes, and tools
A set of operational investments at Meta that scaled the research function's speed and reach, from a Rolling Research program serving multiple teams, to an AI-powered recruiting tool that cut participant sourcing from days to hours, to field research reframed around strategic questions.
My Role
I took on operational challenges across different teams at different points in my career:
🏃♀️While on the Instagram Ads Privacy team, I supported product teams' need for quick, directional qualitative insights without requiring a full standalone study
🤝 While on the Ads Engagement team, I collaborated to streamline participant recruitment reducing dependency on data scientists
🫂While on the Facebook Stories team, I contributed to an empathy field research program and advocated for reframing it around clear business impact to justify the investment
Rolling Research
A lightweight, ongoing user research model where teams have access to short testing sessions on a set cadence. Built prioritization frameworks and onboarded researchers to run it independently.
Faster signal for multiple product teams
Self-Serve Recruiting Tool
Used Claude Code to automate participant sourcing, eliminating dependency on data science for routine recruiting tasks.
3–5 days → 3–4 hours
Cultural Spelunking
Reframed empathy field research around strategic questions, shifting how the team justified and scoped field research investments.
Stronger investment case
Rolling Research
The Rolling Research program gave product teams ongoing access to short qualitative testing sessions on a set cadence, providing faster signal on tactical questions without requiring a full standalone study for every decision. When I took ownership on Facebook Stories I managed the end-to-end operations: vendor relationships, moderator quality, participant recruiting, and coordination across rounds. At its peak the program ran multiple rounds per month serving several teams across Stories Experience and Creation.
When I moved to Instagram I collaborated with colleagues to adapt the model for a new org. My specific contributions included evaluating vendors, building a prioritization checklist that helped the program lead evaluate which incoming projects were a good fit for the rolling research format versus a standalone study, and coaching researchers to continue running the program independently.
Self-Serve Recruiting Tool
The existing recruiting process required a researcher to brief a data scientist on what they needed, wait for the DS to identify the right data table and write a SQL query, and then manually join that output against a consent list of users who had agreed to receive research emails. Depending on the data scientist's availability and prioritization, this took three to five days per study.
Working with four other researchers at Meta, I used Claude Code and Meta's internal AI assistant to automate this workflow. We built a script that identified the correct data table based on research criteria, generated and ran the query, and automatically scrubbed the output against the consent list, bringing the process from days to hours.
Cultural Spelunking
Cultural spelunking trips were designed to build empathy among cross-functional team members through immersion in users' lives across different markets. Over time a question emerged about whether empathy alone was a sufficient return on the significant investment of time and resources. I contributed to the evaluation of the program and advocated for reframing trips around specific product use cases, so that the empathy built during immersion was scoped toward defined questions and connected back to the product development process.
