At Daily Harvest, I drove AI innovations and operations optimization as a machine learning engineer by leveraging large language models and predictive algorithms to enable new capabilities around personalization, unlock insights, streamline inventory and logistics, and build scalable data infrastructure and machine learning tooling.• Engineered a named-entity-recognition (NER) model leveraging PyTorch, Kubernetes, Vertex AI, and a fine-tuned LLM to extract key sensory descriptors from customer feedback, enabling enhanced personalization in product listings.• Designed and implemented a novel machine learning solution to uncover more than 100 latent consumer preferences from 100M+ low-velocity data points, leveraging XGBoost for predictive modeling and Shapley values for insight, boosting content personalization effectiveness across marketing channels.• Architected an automated machine learning pipeline leveraging GPT-4 and LangChain to efficiently generate synthetic product concepts, identifying over 2,500 potential new portfolio opportunities for strategic investment.• Led the company-wide adoption of Notion by creating customized training materials and streamlined workflows that integrated documentation, project tracking, and cross-functional collaboration into a unified platform.• Built an automated data pipeline with Python, Airflow, and DBT to aggregate unstructured text data from non-traditional sources and flag order-related issues with a fine-tuned LLM, increasing Care team efficiency by 8%.• Optimized the allocation of dry ice for customer deliveries based on delivery conditions and a model-informed melt probability, resulting in a 3% reduction in dry ice usage, a 13% decrease in damaged orders, and higher logistic-related SLAs.

• Created and deployed a “Smart Reply” application, leveraging state-of-the-art transformer models, to assist customer service representatives to resolve player issues, decreasing advisors’ handle time and improving customer satisfaction.• Analyzed and modeled user engagement of over 70 million gameplay session to quantify and optimize revenue impact of call center activities. • Enabled SEO experts to improve search rankings by visualizing complex word embeddings and other unstructured metadata in an easy to understand dashboard.• Maintained workflow management server and over 25 automation scripts that deliver regular reports, execute ETL jobs, retrain models, and perform batch model inference.

• Developed and maintained data products and machine learning assets within EA's Worldwide Customer Experience department• Built unsupervised topic modeling report for specific products and services within EA. Results delivered to key stakeholders on a regular cadence.• Orchestrated data collection and annotation to build a novel supervised classification algorithm

• Providing audience segmentation, behaviors insights, and data analysis that drive more effective marketing activities and increase overall revenue• Developed and deployed a recommender system using Python, SQL, and R, which suggested relevant courses to students based on prior behavior with the company• Segmented direct mail audiences using machine learning algorithms based on behavioral data• Delivered insights on corporate retention, informing a strategy that is expected to increase retention approximately 40%

• Leveraged python to collect 40+ million tweets on presidential candidates during the 2018 US presidential election• Assisted Dr. Wendy Moe and Dr. Sarah Oates gather critical data for their working paper "Donald Trump and the 'Oxygen of Publicity'"• Utilized natural language processing (NLP) to assign tweets to unique categories• Optimized data storage and processing in order to feed Tableau visualizations

• Developed an automated workflow using Python and Google Sheets which supplied our legislative team information on bills proposed to specific committees• Researched and advised Senator Steve Waugh on critical bills in the Senate Finance Committee

Some of my freelance work for Senator Waugh included utilizing Python to gather available unfiltered data quickly from Maryland's General Assembly website. Additionally, myself and other statistical experts applied statistical analysis techniques to identify more efficient ways to promote and sponsor legislation.Ultimately, we gathered data that would assist the Senator in not only making smarter choices in terms of legislation and strategy, but also more efficient ones.

Job responsibilities revolved around a new software-as-service product called Social Delta.• Recruited new beta users to test and evaluate SocialDelta functionality and tools• Generated brand awareness among bloggers through product demos• Designed an engaging website in order to increase leads and conversions

During my time at the IRS, I helped develop and deliver cost saving advice to specific departments within the organization. Additionally, I utilized Excel VBA language to expedite the development of reports. My largest project was conduction a Business Performance Review of the CFO division.