I lead the data organization, a 90+ person group of data scientists and engineers covering:• product analytics: influencing decision-making through opportunity sizing, metrics/dashboard design, ad hoc and strategic analyses, and experimentation design/analysis/interpretation.• analytics engineering: delivering standard analytics-ready, high-quality data assets (i.e., tables, data dictionaries) for core data and process domains.• data products: developing production-ready machine learning and deep learning-based systems that directly power customer experiences and drive internal efficiency.• data governance: establishing the processes and policies used to ensure the availability, usability, integrity, and security of data.• data platforms: providing the fundamental systems and infrastructure used by all data applications, such as data warehousing, data operations, resource provisioning, and workflow management.• data services: providing core online services that enable the effective collection, processing, and utilization of data by developing libraries and internal APIs that connect the data stack with online web and native applications.Udemy is the world's largest marketplace for teaching and learning online. We're tackling a broad range of hard technical challenges, including (but not limited to): pricing optimization, course quality measurement, understanding LTV drivers, content discovery, knowledge graph construction, experimentation analysis, and scaling batch and real-time data infrastructure.Interested in joining the team? Send me an email at chuong.do@udemy.com!

Built and led Coursera's end-to-end data organization from scratch to ~30 in size. Had the privilege to work with an amazing cross-disciplinary group of software engineers and data scientists spanning:• decision science (metrics, reporting, experiment design, statistical analysis, forecasting)• data products (modeling, algorithms, machine learning, search/recommendations, "next generation" technologies)• business intelligence (data governance, curation, standardization)• data infrastructure (data warehousing, data access and visualization, A/B experimentation, surveys/instrumentation, machine learning infrastructure, personalization, research, email infrastructure, data science productivity)Wore multiple hats at the company, including hands-on engineering and data analysis, business/product strategy, data infrastructure/product/science roadmap, growth analytics, HR/compensation analysis, etc. Partnered cross-functionally with nearly every team in the company.Highlights include:• Independently built data function from the ground up• Drove vision/strategy for all aspects of data, including long-term infra investments• Guided the org through numerous structural changes in response to evolving company needs and team dynamics• Hired, mentored, and retained a talented team, including many early employees• Ran Coursera's first internal employee survey• Built Coursera's first user growth model and fostered a data-driven culture• Conducted analysis that drove shift away from pure self-paced models of instruction• Invented frameworks for navigating user/revenue growth and free/paying user trade-offs• Developed and drove company-wide adoption of a rational approach to employee equity allocationAt one point or another, directly managed data infra engineering, data products, and data science teams. Simultaneously managed ICs, managers, and managers of managers.Previous roles:• Software Engineering Manager, Analytics• Software Engineer, Analytics

Developed core machine learning algorithms underlying the Ancestry Composition feature, including:• a haplotype graph-based method for large-scale computational out-of-sample phasing using extremely large reference panels (over 100,000 individuals)• a multi-class support vector classifier for ancestral classification of haplotype segments• a dynamic Bayesian network model for simultaneous local ancestry deconvolution and phasing error correction• a regularized isotonic regression-based procedure for recalibration of classification reliability estimatesDeveloped a highly parallelized and scalable architecture for large-scale statistical genetic analysis, including:• a volume-based abstraction for efficient storage and manipulation of dense imputed genotype data• a distributed, multi-threaded system for efficiently running association tests across thousands of phenotypes and millions of markers, based on hundreds of thousands of sample points on a 100-node cluster.Contributed to multiple genome-wide association studies using 23andMe's customer database, including Parkinson disease, hypothyroidism, breast size, myopia, cilantro preference, and assorted medical conditions.Developed theoretical models for SNP-based and family history-based disease risk prediction.Contributed to regulatory compliance efforts.