Fia Finance (React Native app aimed at helping couples manage their finances — in development)• Built React Native app atop AWS Amplify authentication and back-end services• Used Lambda functions to integrate with Plaid API to associate financial accounts with the app• Created data model across users, financial accounts, and transactions using GraphQL• Enabled users to link their accounts with their partner’s via a QR code NFTeeOff (Blockchain-based Fantasy Golf web app using ERC-721 non-fungible player and gear tokens)• Built full-stack React web app with backend, authentication, and deployment managed by AWS (GraphQL schema atop Dynamo database; S3 for image storage)• Created Solidity contracts responsible for token minting, a marketplace for trading, and an auction house• Automated the generation of new token auctions using a Python script that uses user base size to dictate rate• Integrated with real-time sports data API to enable draft functionality and scoring• Deployed dApp via Truffle on Rinkeby test network (now defunct)Fantaseasons (React and React Native Historical Fantasy Football App)• Built full-stack MERN web app that lets users create leagues, draft players from past seasons in real-time (using socket.io), and manage their teams throughout the season• Deployed Python-based governor script randomizes weekly results for season-long competitions• Created companion mobile app with management features (available on both Google and Apple App Stores)Covet Real Estate (Luxury Real Estate App)• Scraped multiple listing services to find properties across the United States selling for more than $10 million• Integrated Firebase back-end and authentication with Swift-based UI to build an iOS app to display listings• Built functionality enabling the user to save and share their favorites or discover new homes using the map, sortable list, or roulette (randomizer) feature

• Spearheaded project to generate foundational representations of users to enable similarity and propensity modeling across multiple data streams, including web browsing activity, financial transactions, and credit report data• Increased email click-through rate and revenue by more effectively targeting CRM campaigns to users whose vector representations contained signals indicating greater interest in particular verticals• Launched an effort to simplify and expedite the personal loan qualification process. Applied tree-based models to predict the expected probability of approval based on questionnaire responses in real-time• Increased conversion rate and revenue on multiple pages by optimizing the ordering of credit card products based on users’ predicted financial strength• Oversaw a junior data scientist in labeling raw transaction data and training an appropriate neural network architecture to recognize personally identifiable information in merchant strings that did not follow grammatical rules (making existing NER models ineffective)• Supervised an effort to predict the time-to-transaction within each business vertical using a recurrent neural network model• Ideated and led one winning hackathon team (integrating cryptocurrency wallets and exchanges in existing net worth tracking tool) and one runner-up (Chrome extension that recommended ideal card to use on cart contents based on potential rewards)

• Managed a long-term cross-departmental effort to implement a computer-vision-based inspection system at the fulfillment center; projected cost savings of over $500,000 per year• Worked closely with infrastructure developers to integrate cameras into the existing Raspberry Pi network that governed fulfillment center operations; built out software to capture and transmit images• Retrained numerous alternative object-detection models and worked with quality assurance experts to determine the optimal trade-off between inference accuracy and latency to enhance operations• Built training data set for proprietary pill detection model by creating and overseeing a Mechanical Turk task and developed a script to process and validate results• Identified the most common product combinations by applying a triadic distance metric to vectors built on user purchasing trends; served to validate the idea of pre-made packs for specific health goals• Developed a prototype iOS app in Swift that collected wearable data (e.g., heart rate, activity modes, etc.) for subsequent potential modeling efforts; continued iterating based on feedback from stakeholders• Oversaw a junior data scientist in coding an algorithm to optimally assign orders across multiple processing machines while satisfying FDA allergen regulations and minimizing the impact on production output; deployed Flask app to expose results to fulfillment center staff for daily planning

Evaluating Spatial Risk of Anthrax Exposure in Plains Zebra from Etosha National Park, Namibia• Cleaned and parsed nearly 750,000 empirical GPS points from zebra and springbok; simulated and analyzed over 97 million movement points (9.65 GB)• Generated a predictive spatial map of anthrax suitability using maximum entropy algorithm• Applied hidden Markov modeling method to classify GPS time-series data in R• Developed distance-weighted sampling algorithm in Python that reduced runtime by 68%• Built logistic regression models to evaluate predictors of zebra habitat selection in R• Created a mechanistic agent-based simulation of movement to approximate contact processDefining Ecological Niche of Zika Virus and Projecting Global Pandemic Potential• Applied several machine learning methods (GLM, Boosted Regression Trees, Random Forest, and Artificial Neural Nets, among others) to delineate the ecological niche of Zika Virus in R• Used Principal Component Analysis to compare niche to a related viral pathogen (Dengue)• Simulated epidemiological trajectories of Zika infections across the continental United States• Parallelized stochastic simulations on Linux-based high performance computing clusterRefining Epidemic Disease Transmission Model Formulations: Ebola Virus as Case Study• Proposed alternative differential and difference equation sets to account for heterogeneity• Used Markov Chain Monte Carlo (MCMC) methods to fit new models to Ebola incidence data• While computationally more demanding, the discrete formulation improved model fit (based on negative log likelihood) by nearly 7% compared to the standard SEIR model

• Extracted genetic material from scat, tissue, or hair samples obtained from researchers around the world• Utilized various lab techniques to amplify, clean, and sequence the extracted DNA• Analyzed mitochondrial and nuclear DNA sequences to determine species identities and population structure in order to inform conservation decisions• Published the first complete mitochondrial genome of the jaguar (Panthera onca)

· Assisted in the logistics and set up of multiple long-term experiments concerning aquatic plant communities facing simulated drought and high nutrient loads · Collected abundance and diversity data on macro-invertebrates and amphibians from artificial and natural ponds throughout Missouri · Injected fluorescent elastomer into amphibian appendages as part of an ongoing mark-recapture study of Grey Tree and Cricket Frogs

· Aided in the planning and implementation of Swamp Wallaby census study (Advisor: Daniel Ramp) · Deployed camera traps throughout two National Parks and re-collected for analysis on a weekly basis · Developed mathematical model to estimate population density and abundance of species without individual markings that move within distinct home ranges using camera trap data