Tech Lead for Inventory data science team

• Established simulation and forecasting initiatives as second data science hire• Created logistics simulation platform from scratch to support business decision‐making and to serveas development harness for routing software• Used Ray and Anyscale for large‐scale simulations – first to use these technologies at Veho• Conducted simulation analysis to estimate upper bound of gains from demand forecast which savedmonths of live experimentation• Facilitated multi‐million dollar savings initiative by providing analysis of trade‐offs between routingefficiency costs against warehouse efficiency gains• Designed data scientist technical screen and conducted interviews of all candidates

• Led ads safety data science team which included myself on ad policy enforcement and two directreports on traffic quality and brand safety• Implemented four ad policy enforcement models in GCP, more than quadrupling the number of policyviolations detected• Developed pipeline for faster model deployment and routine retraining that was on track to deploya dozen more models by the end of 2022• Mentored data scientists from other teams on technical writing, presenting, and management• Directed implementation of measurement systems for brand safety and traffic quality

• Contributed to software that implemented statistical and machine learning models for forecasting order volume and courier throughput in each neighborhood across the country.• Contributed to Python package for power analysis through simulation speeding up experimental design and ensuring valid designs were used by all data scientists in delivery division• Designed complex experiments for cases where network effects precluded simple a/b tests and evaluated results using hierarchical regression or cluster‐robust standard errors as appropriate

• Built an LSTM neural network forecasting model that issued weekly predictions for hundreds of thousands of customers to guide individualized marketing and high touch outreach• Built tools for other data scientists to retrieve census data and created a pipeline to store census block group data related to our business locations• Trained data scientists on Git, Bitbucket and code review process

• Contributed to software that implemented statistical and machine learning models for forecastingorder volume and courier throughput in each neighborhood across the country• Introduced improved methods for combining ensemble of forecasts that saved over $1M in laborcosts by avoiding over‐scheduling. Also improved delivery times and courier satisfaction by avoidingunderscheduling.• Devised novel procedure for modifying prediction intervals that improved sharpness without degradingcalibration• Built data pipeline that transformed weather data for input to forecasting models and combined itwith weather forecasts at time of prediction

• Hired and managed three full‐time staff who administered the Undergraduate Admissions Office’sCRM used by 20 campuses and 13 colleges• Built a machine learning model for marketing segmentation of prospective students• Implemented marketing automation and content strategy, reducing event marketing workload by79%• Resolved contentious debate over event schedule with new schedule built using frequent patternanalysis of event attendance data. This enabled us to make more effective marketing investmentsand increase event registrations by 7%