Emily Vu, PhD

- Led global team of 12 data scientists across 4 regions as Scrum Technical Lead to develop central Bayesian hierarchical regression model to quantify key sales drivers for brands across 6 categories - Spearheaded and transformed global collaboration within Data Science organization by establishing central codebase, enforcing GitHub best practices, and promoting organized centralized documentation, which accelerated model innovation and release - Managed code development using Scrum framework, facilitating Program Increment Planning, sprint planning, and stand-ups to optimize team performance - Cultivated supportive environment that encouraged open communication and collaboration, enabling improved team dynamics, individual growth, and a high-performance culture - Aligned expectations with key stakeholders through regular updates and technical discussions, driving clarity on methodology and deliverables - Validated, scaled, and operationalized model for 8 brands across 6 categories and 4 regions, increasing modeling frequency to quarterly and enabling explainability of insights Individual Technical Contributions: - Organized hackathon to migrate initial notebook implementation to modularized pipeline, launching POC model w/in 8 weeks using PyMC package to support single-hierarchy-level data - Implemented and verified advanced multi-hierarchical model supporting three levels enabling deeper insights across various granularities - Reduced model size by 70% to enable efficient execution of large, multi-hierarchical models - Developed synthetic data pipeline to verify model accuracy, improving media and non-media priors calculations for better model tuning - Received many Power of You (POY) awards for excellence in execution, leadership, and teamwork, including the 2024 Global D&A POY award for disrupting decision-making processes

- Within Global Financial Services Vertical, engaged with global financial customers to complete 6-week assessment of applications, deliver proposed architectures, and document risks/dependencies, weekly status, financial reports, and cloud migration cost estimations - Supported the launch of the Associate Engagement Manager Community of Practice Chapter to provide catered foundational knowledge and support for consultants within Amazon Web Services, Inc. Professional Services - Led cross-functional teams in implementing AWS solutions, ensuring successful project delivery - Collaborated with clients to understand their business needs and translated requirements into effective AWS strategies. - Managed stakeholder relationships, fostering strong communication and ensuring client satisfaction. - Provided technical guidance and support to clients, optimizing their use of AWS services. - Contributed to business development efforts, identifying opportunities for upselling and expanding service offerings. - Monitored project timelines, budgets, and deliverables to meet and exceed customer expectations.

- Worked with Corporate Function Information Technology Data Science Retail Team - Developed a user-friendly, modularized store clustering notebook that performs store clustering based on product dollar share performance and is usable for few hundred members within Friends of Data Science (FODS) community on category and customer teams - Produced Power BI template to visualize clusters and customer demographic information per cluster to gain insights on product performance based on business unit, product attributes, and neighborhood demographics - Reapplied implicit collaborative filtering model on standardized Nielsen Disaggregated Household Panel data to gain further insights on inherent biases within datasets on various business use cases, particularly Target retailer fem care assortment analysis - Developed methodology to easily identify biases within subsets of Nielsen Disaggregated Household Panel data based on categorical percent penetration and item dollar-share distribution - Implemented a machine learning pipeline to solve the cold-start item recommender problem and enable newly launched item recommender capabilities using Nielsen Disaggregated Household Panel data

- Implemented Monte Carlo algorithm Local Realization Preserving in production-level code Mercury. - Expanded stochastic transport capability to handle material mixing beyond binary. - Produced benchmark results using Monte Carlo algorithms Chord Length Sampling and Local Realization Preserving for one-dimensional, binary, Markovian-mixed media and conducted accuracy comparisons

Radiation Effects Theory Department - Implemented a limited-memory framework to enable higher-moment computations (probability density functions) for particle transport in 1D stochastic media using Conditional Point Sampling. - Conducted accuracy comparisons and parametric studies on computer memory footprint, computational runtime, and accuracy

-Implemented limited memory techniques in new method for radiation transport in stochastic media. -Conducted numerical studies to understand cost savings between runtime, accuracy, and computer memory requirements. -Findings reported in ANS transactions paper (reduced memory implementation) and ANS Mathematics and Computations Division conference paper (amnesia radius memory implementation).

-Implemented an extension of new method for radiation transport in stochastic media to calculate the variance and uncertainty of the variance of mean results due to random material mixing (parametric variance). -Conducted numerical studies to understand parametric variance behavior due to various free parameters. -Findings reported in ANS Mathematics and Computations Division conference paper.

-Used Python to implement a new Monte Carlo radiation transport method in stochastic (Markovian-mixed) media leveraging Woodcock tracking for particle simulation. -Demonstrated the basic method and properties in a 1D implementation using two differing derived conditional probability functions in implementation. -Showed that of the two implementations, one is capable of producing results as accurate as many established approximate methods and the other is capable of producing results with no bias error for special case of 1D Markovian-mixed media. -Findings recorded in American Nuclear Society (ANS) conference paper.

-Used C++ to implement Woodcock tracking and biased Woodcock tracking in one-dimensional, Markovian-mixed media -Benchmarked results against analytic attenuation results

-Course reader in UC Berkeley’s nuclear engineering department -Grade homework assignments for Numerical Simulations in Transport Equation (NE155) and Nuclear Power Engineering (NE161)

Project Title: An Angle-Informed Hybrid Method for Neutron Transport (CADIS-Omega) -Characterized a hybrid method (CADIS-Omega) developed specifically for highly anisotropic radiation transport problems. -Used hybrid method to model neutron transport in a dry cask.

Project Title: Compact Integral Effects Test (CIET) -Independently learned SNAP and TRACE computer language (used by the NRC) to modeled Compact Integral Effects Test (CIET) facility. -Conducted steady-state tests through TRACE and validated model using CIET experimental data. -Conducted code-to-code bench-marking against RELAP5-3D data to further verify TRACE and RELAP5-3D evaluation models. -Conducted safety analysis experiments on CIET facility for Flouride-Salt High-Temperature Reactor (FHR)

-Worked with Probabilistic Risk Assessment group on identifying systems and components with appreciable fire risk for internal events -Computed importance of each system and component using winNUPRA based on risk achievement worth values and risk reduction worth values -Identified fire risk management actions for necessary systems and components based on cutsets identified in output files -Created ePSA database to organize peer reviews and Findings & Observations reported in Gap Analysis for various PRA models

-Conducted radiological assessment of lead fast reactor (LFR). Quantified radiological footprint where possible and provided roadmap for the development of the LFR based on previous work on PWRs such as AP1000, AP600, SMRs, and APWRs. Current tools and methodologies were reviewed, and potential technical gaps were identified. Findings documented in a letter (LTR). -Assessed changes of NRC's Regulatory Guide 3.54 by implementing decay heat calculation methods in Matlab for Regulatory Guide 3.54 Revision 1, Regulatory Guide 3.54 Revision 1, and ORIGEN-ARP (Westinghouse's current tool for calculating decay heat). Verified and validated code. Assessed impact of Regulatory Guide 3.54 changes on CaskWorks (uses ORIGEN-ARP method) by analyzing accuracy and precision of decay heat calculations for each method based on decay heat measurements of PWR and BWR fuel assemblies. Measurement-to-Calculation ratio for different accumulated burnups and fuel enrichments were also observed. Results and analysis were documented in a calculation note. -Updated Ex-Vessel Neutron Dosimetry (EVND) database for 3-loop and 4-loop plants, updated calculations, plots, and calculation notes. Verified and validated code used for documentation. -Provided over 400 plots and figures for Watts Bar Unit 2 WIN-CISE report analyzing vanadium detector response to changes within reactor -Ran Low Power Physics Test (LPPT) on Intermediate Range Digital Reactivity Computer (IR-DRC) system for Diablo Canyon Unit 1 Cycle 21 outage -Began formulating RHOPro (IR-DRC) User's Manual based off of Advanced Digital Reactivity Computer (A-DRC) and Subcritical Rod Worth Measurement Data Analysis System (SRWM-DAS) User's Manuals.

-Worked part-time (20 hours/week) during school year -Graded assignments, quizzes, tests, and exams for subjects such as Algebra I, Algebra II, Geometry, Pre-Calculus, and Health. -Encouraged and guided students from all 50 states and worldwide in developing mathematics and health skills.

-Conducted experiments and tests to analyze bifurcation in structures. -Utilized Mathematica programming language to analyze data and write comprehensive reports on findings