Computer vision, natural language processing
Quantitative researcher with expertise in data analysis, statistics, and machine learning. Over 10 years of experience in planning and execution of data-driven projects; including the development of data collection and analytics pipelines, and clearly communicating insights to stakeholders.
- Research partner for AAA PC, console, and mobile games.
- Conducted surveys and A/B tests to extract usability insights from real-world users.
- Worked in tandem with designers to apply iterative research to concepts.
- Programmed internal tools (Python, C#) to collect, analyze, and visualize datasets.
- Led client meetings to scope project requirements, develop research plans, and present reports.
- Developed quantitative onboarding program for training new researchers.
- Designed and implemented algorithms for extracting novel spatial metrics from HoloLens 2 sensors.
- Constructed data pipeline to collect, clean, and analyze hardware sensor data.
- Performed exploratory analysis to evaluate the usability of new metrics in different production contexts.
- Presented insights to the data science team regarding the value and potential pitfalls of implementing the spatial metrics in production.
- Extensive training in statistics, which was applied to the analysis of over 16 published experiments.
- Constructed data pipelines, from data collection, transformation/aggregation, to data analysis.
- Developed algorithms to extract model features from raw multidimensional sensor data.
- Created Python dashboards to aid cleaning of structured data. Reduced data cleaning time by 90%.
- Programmed open-source data collection platform to collect experiment data across 9 projects, resulting in 50% faster project completion. Link: GitHub/cognitive-battery
- Created open-source Python package to analyze time-series data. Link: GitHub/sensormotion
- Modelled time-series data using multilevel regression and analysis of variance.
- Published results in 7 peer-reviewed journals and gave presentations at 11 research conferences.
- Employed experiments and quantitative methods to help instructors evaluate new teaching methods.
- Analyzed A/B test involving 6000 online students to identify factors that affected engagement.
- Conducted experiments to determine whether the benefits of a novel visualization technique outweighed the cost. Results saved over $2000 in future production cost.
- Used statistical methods to streamline a multi-institution survey. Reduced total survey length by 28%.
- Developed a webcam-based eye tracker to predict real-time eye gaze location.
- Programmed data collection pipeline to collect and transform 50k webcam images.
- Built deep learning computer vision architecture, using PyTorch, to model webcam images.
- Achieved a final eye tracking error of 1cm, rivalling more expensive infrared-based approaches.
- Published novel method to measure outdoor walking kinematics using smartphone sensors.
- Technical lead on cross-functional, collaborative research project with the University of Bristol.
- Designed algorithms to compute walking metrics from 60 million data points.
- Used multilevel regression to model the relationship between walking kinematics and cognitive function.
- Method was faster and cheaper than traditional techniques, saving over $7000 in equipment cost.
- Incorporated insights from usability testing into the design of 3 new site features.
Cognitive Science, minor in Quantitative Methods