Industry · Tanya Wen, Ph.D.

Work Experience

Aug 2025 –

Data Scientist

Meta, Family of Apps

Conducted a 4-part benchmarking analysis of internal and external messaging apps using production data and lab experiments to understand engagement, reliability, render quality, and user experience across 6 media types.
Identified opportunities to personalize the ML model for media uploads (optimizing reliability, latency, and visual quality tradeoffs); ran simulations to size impact across user cohorts and inform experimentation.
Analyzed video prefetching across Facebook and Instagram; reviewed 50+ historical experiments to identify 7 key optimization levers and ran 10+ A/B tests to improve video buffering and egress usage.
Root caused 4 major regressions in media reliability and 6 common error codes; identified 3 logging bugs and partnered with engineers to land fixes and reprioritize the team roadmap.
Built an automatic metric collection scheduler and 5 dataswarm pipelines powering the Media Foundation team's experiment tracking dashboard with regression monitoring.
Built 8 analytics agent recipes with core table context, SQL query patterns, decision logic, and validation checks — accelerating regression investigation from days to ~30 minutes and automating weekly reporting.
Supported 60+ A/B test reviews through office hours covering metric selection, results interpretation, and launch readiness; identified Media Messaging improvements driving +30M daily sends and +5M daily sessions.

Mar 2024 – Aug 2025

Applied Perception Scientist

Meta Reality Labs Research – Audio (contract via Magnit)

Led experimental design, analysis, and documentation for four KPI metrics across 5 studies to A/B test audio personalization on the Quest 3 VR headset, informing the team's technology roadmap. Resulted in two publications and open-source contributions.
Led a UXR study evaluating user value of a newly developed audio renderer for avatar-based mixed reality conversations, using Bradley-Terry ranking models and qualitative thematic analysis.
Co-led onsite user studies on speech intelligibility for Meta Ray-Ban glasses, measuring psychophysics thresholds and qualitative ratings — contributing to the Conversation Focus feature announced at Connect 2025.
Built a web-based data collection infrastructure using Qualtrics and Amazon S3, increasing the team's data collection efficiency by at least 30× and reducing cost by ~80%. Demoed to stakeholders across XFN teams; became a primary data collection method for the team.

Oct 2022 – Mar 2024

Research Scientist

Naval Health Research Center (contract via Leidos)

Led EEG analysis in 50+ traumatic brain injury patients and healthy controls, using ERP analysis, k-means clustering, and exploratory data analysis to identify biomarkers. Results included in annual report and conference presentations.
Designed and programmed four virtual reality tasks in Unity/C# with hardware integration (light sensor, sound sensor, IMU) and timestamped event streaming to Tobii eye-tracking glasses for multi-device synchronization.
Led data analysis for a VR experiment measuring gaze stability, fixations, and saccades using a wearable eye-tracker and infrared camera; developed custom analysis scripts using OpenCV and pre-trained neural networks.

Skills

Quantitative

A/B testing Regression ANOVA t-test Linear mixed effects models Exploratory data analysis Supervised learning Causal inference Forecasting Opportunity sizing

User Research

Experimental design Metrics design Virtual reality Eye-tracking User interviews Online studies Thematic analysis Sentiment analysis Surveys IRB application

Programming

Python SQL R MATLAB JavaScript Unity / C# Git / GitHub Data pipelines NumPy Pandas Scikit-learn

Resume

Download Resume