(2023-2024) Designed, coded (as team lead), and product managed the first Chartboost DSP auction-level market price prediction model for first price auctions- Functioned as both product manager and technical lead: wrote requirements and product design documents- Got broad stakeholder buy-in via concept explanations, interactive demonstrations, and clear visualizations of the opportunity and solution functionality(2023) Partnered with Data Engineering and DevOps to stand up orchestration resource for control and optimization products that exemplified industry-standard practices- Designed and led implementation of tunable explore-exploit use case for new advertiser onboarding (addressing “cold start” problem)- Led team that stood up Airflow environments (staging and production) with CI/CD via Github actions and wrote maintainable code via introduction of Jinja query templating(2022) Architected, coded (as team lead), product managed, and shipped on schedule a budget automation solution that met all requirements and controlled over 50% of DSP advertiser spend- Blended fundamental control principles (a discrete-state controller) with more sophisticated or novel methods, including: - - Active learning of advertiser specific bid-multiplier boundaries - - Pacing system based on continuous measurement of the price elasticity of installs and in-app purchase revenue- Specified the comprehensive suite of monitoring, alerting, and fallback logic measures- Partnered with platform engineers to expand the uses of Redis and Kafka as appropriate for our demand information layer(2021) Designed, shipped, and evaluated an early campaign control and optimization approach using a PID controller- Wrote core logic in Python- Got broad stakeholder buy-in by delivering detailed, interactive spreadsheet model of system operation

Brought on DevSecOps contractors for infrastructure work and liaised with external security auditorsSet up basic data pipelining and outcome definitionsBuilt initial clinician performance reporting workflow; used empirical Bayes technique to account for varying ratings counts. Built Excel tool for clinical ops to use Erlang C formula to manage support ticket staffing

Twenty® was formed out of a need for today’s plugged-in generation to spend more time connecting with their friends in real life instead of through a screen. Our goal is to drive more human connection by making it easier than ever to make plans and create new opportunities for friends to meet up. • Responsible for cultivating company-wide culture and best practices for data-driven product design, decision-making, and knowledge sharing• Set long and short term vision for data strategy, team contributions, and individuals’ career growth• Hired and lead team of 4 data scientists/engineers (2 FT/2 contract)• Incorporated social network analysis techniques and machine learning (multi-armed bandit) to product features and growth efforts• Developed method for detecting close-contact encounters (detection of friends hanging out)

Acquired by Twenty in 2019. Initial data science hire; built systems for internal reporting, product analytics, and rigorous decision making supporting growth of the Mappen app from 0 to 2 million users.Contributions:• Partnered with product and engineering to refine thesis and performance criteria for every major feature in Mappen• Developed and justified rules for rate-limiting push notifications• Used graph centrality measures to identify subpopulations of user base and likely influential users• Built out simple yet thorough internal reporting system • Spec'd and built data warehouse in MySQL of analytics-focused aggregate tables for product managers and marketers; built pipelines using MySQL Event Scheduler and crontab

Owner and subject matter expert for experimental design across all marketing channels; developed and implemented A/B testing and reporting framework.

Advised and helped plan initial data capture and analysis approach for mobile app aimed at improving medication access and quality

Designed and implemented comprehensive Business Intelligence workflow resulting in reproducible, efficient, financially-impactful analysis work supporting all company teams.Championed adoption of modern data science tools (eg, git-tracked Python/R, literate programming, AWS tools, CLI use).Team lead for 4 analysts; mentor in areas of statistical practice (sampling, data transformation, results interpretation), project selection, and working with large, less-structured data collections.Developed and directed Chartboost network health program, incorporating ongoing feature engineering, EDA, text pattern analysis, and process validation (qualitative and quantitative false-positive management).Prioritized and directed analyses on new products, supply and demand matching, revenue forecasting.

Client-facing data engineer, accountable for building and maintaining data pipelines relied on by hundreds of RTB buyers and sellers.Optimized internal hardware allocation, client onboarding, account management, and billing processes.Languages used: Pig, Scala, R

Small Business Products and Services Intern within Global Network Services. Developed strategy for introduction of premium small business card product; led development of internal website; worked with external agency to develop creative asset library.

Led cross-agency evaluation of a $40M electronic health surveillance system for former Soviet states; findings presented to Assistant Secretary of Defense for Health Affairs.Advised allocation of a $3 million competitive research budget.Coordinated respiratory disease data reporting from laboratory network spanning 5 continents.

Developed military mental health dashboard metrics presented to Deputy Assistant Secretaries of Defense.Developed diagnostic code signature with optimal sensitivity and specificity for surveillance of potential smallpox vaccine-associated myopericarditis cases.Constructed and interpreted logistic and log-linear models of health outcomes.Advised academic researchers on use of Defense Medical Surveillance System for cohort and nested case-control study designs

Estimated sensitivity and predictive characteristics of Pre- and Post-Deployment Mental Health Screening Tools.Analyzed seasonality of deployment histories to "rule in" potential locations of P. vivax infection.

Used ANOVA, nonparametric methods to inform optimization of vaccine production process.Recommended use of better-powered exact tests rather than GLM-derived ID50s to compare infectivity of P. yoelii sporozoites undergoing candidate preservation methods. Constructed creative visualization to demonstrate effect of temperature variance on mosquito infection rates.

Matched-pairs trial of microfinance intervention for HIV incidence and gender-based violence. Recorded and facilitated focus group discussions, assisted with data cleaning and results interpretation, shadowed professionals building out HIV testing & counseling capacity in rural South Africa.