Head of Consumer Product Data Science• Product Management – Set roadmap and defined requirements forPersonalization• ML Engineering and Ops – Created a generalized template for Data Scienceas a microservice• Infrastructure and SRE – Defined SLAs for Data Science as a Service.Architected and currently supporting many microservices that are on the criticalpath for our consumers• Third party integrations – Evaluated build vs buy and established relationshipswith multiple 3rd party providers to enable team to scaleEnabling team members• Data Engineering - Created a geospatial processing and visualization engine toenable entire Analytics team.Hiring• Scaled analytics team from 4 in 2019 to 14 (so far) in 2021Strategy• Defined process and created company OKRs for 2021

Impact• Drove $49.7MM in GMV with improved Recommendations engine in 2020.• Drove $55.1MM in incremental GMV with ETA microservice 2020.Process Improvements• Upgraded both ETA and Recommendations to scale efficiently with growinguser base and handle 11X increase in daily traffic.• For the Recommendations engine, improved provisioning strategy andbespoke partitioning led to a 643% speed up and reduce costs by factor of 4.

• First Data Scientist in• Recommendations Engine that drives 10.3% of company GMV: Designed, secured buy-in for, and built an in-house Recommendations engine. Stood up end to end architecture to support all Recommendations on site combining EMR, ECS, Elasticache to surface multiple relevance models with sub 10ms latency.• Estimated Time of Arrival model that drove a 5.5% conversion increase: Designed and secured buy-in for, and built an ETAs model as a service. Stood up end to end architecture exposing models as a REST endpoint by integrating Elasticache, ECS, SNS, EMR and implementing a nginx+gunicorn+Flask microservice. Responsible for a 5.5% conversion increase (a ~16.7% improvement from baseline)• ML Ops Infrastructure work: Created a real time feature store with features generated by Spark and persisted in Redis to feed multiple systems at <2ms latency.• Strategic work: Provided econometrics guidance to the C-suite to define company strategy and prioritize which markets to expand into.• Defined our retailer ranking system• NLP and NER: Built out an entity extraction system to augment thousands of POS integrations

Built programmatic advertising platform and DSP for use by advertisers that led to an acquisitionMajor outputs:Cross Device User Identification:• A novel graph processing algorithm research and implementation unifying factorization techniques with graph learning and cutting algorithms to perform on device graph of 2+BN nodes and 1+TN edges.• Drove an improvement in positive prediction accuracy (PPV) of 45% over previously existing solutionPerformance Optimization: • Design, implementation, and testing on the predictive engine that drives advertising performance KPIs on our DSP