Building cradle-to-grave automation for P&C insurance claims. Ensuring alignment with human expertise.

Automated derivation of analytics from previously untouched Non-Equity exchange data using Python, Kafka, proprietary technologies. Built internal apps to automate the creation of futures and options tickers.Resigned after four months to build at a startup.

Built and scaled a community manager for student orgs at Vanderbilt. Grew adoption to 450+ orgs (80%) with 90% student adoption and 210k+ lifetime engagements.Added QR-code checkin, analytics, roster management, link shortening. Designed architecture and UX from scratch and rebuilt platform into a multi-tenant solution for future expansion.Used Node.JS, Python, Typescript, MongoDB, React, GraphQL, Next.JS, Firebase, AWS Elastic Beanstalk, GCP. Paused development due to lack of administrative support.

Built a platform for experimental research (psychology, misinformation, marketing) into how people engage with different types of content on social media.Created hyper-realistic simulations of Instagram and YouTube with full-featured interaction, navigation, and stateful persistence. Built an engine to dynamically populate website content based on experimental conditions and between-subjects randomization. Dynamic generation of client-side GraphQL data schemas and content, including client-side data query engine. Realtime browser event streaming platform for granular reporting on all user actions within platform simulations. Built in Node.JS, TypeScript, Next.JS, React, MongoDB, GraphQL, Kafka.Paused to focus on other projects.

Collected and applied big data in a project examining the library-level effects of book bannings.Scraped 55,000 online school libraries, collecting over 6TB of data and over 4.5 billion book listings. Wrote custom scrapers and batching server, deployed container swarm (Python, NodeJS, MongoDB, AWS ECS, Docker). Designed and implemented data pipeline in Spark for processing and analysis. Developed book-ISBN clustering algorithm for data deduplication, reducing dataset 800x from 4 billion entries to 5 million (PySpark, AWS Athena, AWS Glue).

Held various roles in pro-bono nonprofit advisory group. Built a custom ATS platform using ML to automatically redact candidate PII and reduce unconscious bias, a nation-wide first in higher education.Identified 9 nonprofit clients including organization's first NGO. Led an LMS vendor evaluation and implementation for arts nonprofit's transition to virtual learning. Redesigned membership structure for an association of free and charitable clinics.

Researched CHIPS act, international AI talent competition between US and China, and various other topics. Contributed to cybersecurity reports for the Gates Forum and Institute for Genocide and Human Rights Studies. Performed quantitative analysis of 230K Tweets using R to visualize Russian misinformation in Latin American countries re: Ukraine War. Co-wrote weekly tech/policy newsletter.

Researched socio-political inequality in America through the collection and analysis of digital media.Collected every Yelp restaurant review from 2019-2020 (2M+) using custom web scrapers deployed in a container swarm (Python, Node.JS, Cheerio, MongoDB, Docker, Azure). Conducted sentiment analysis using NLTK, SpaCy to observe ratings drop-off in Chinese restaurants during COVID-19.Scraped all IMDb data and visualized key TV themes by decade. Scraped all Fox News content from YouTube and Twitter over a two month period with 30-minute resolution and built an analytics dashboard to visualize content performance by keywords and theme.

Worked on a team building a social listening tool and trend prediction engine for a Fortune 100 cosmetic company. Migrated storage structure for 35M record data pipeline using Pyspark, AWS (Glue, Athena).Developed ETL pipeline for scraping TikTok data with Puppeteer and reduced ingestion failures by 95% system-wide through the implementation of datacenter proxies. Added automated failure alerting to Teams. Created proof of concept "influencer discovery" feature using NLP and graph density analysis with SpaCy, NLTK, and NetworkX. Identified dozens of new themes to track and reported 10+ new target influencers in weekly client report.

Built NLP tools for automated review of insurance legal documents. Created pipelines for extracting key legal information from medical invoices using Tesseract, SpaCy, NLTK, Regex. Trained short-text LDA topic analysis models on Twitter data using Gensim and clustered corporate Twitter accounts based on thematic analysis of their posts.

Developed minigames and cosmetic updates for the world's largest Minecraft server at the time with 40k max daily active users and playerbase of 5 million.Singlehandedly developed three minigames (Bomb Lobbers, Monster Maze, Evolution 2015) in Java, member of development team for four more (SkyWars, Christmas Chaos, Halloween Horror, Slime Suckers).