This past year, my focus has been on using LLMs and related technologies in developing complex multi-agent systems to assist in research and patient recruitment. Was a primary back end contributor to recently debuted recruitment software "Muse", architecting and implementing many key features of the application such as a low-code configuration and execution scheme for multi-step AI assistants, and automating IRB-like validation/evaluation in produced materials.More broadly over the past 3 years, I've apply primarily TypeScript (React, NestJS) and Python in creating applications supporting research & patient recruitment for clinical trials, aiding Formation's mission to bring new treatments to patients faster and more efficiently.

Automated, optimized, and expanded upon system for transforming vast amounts of hospital data from disparate sources into the OMOP Common Data Model, facilitating the data's use in MSK's cancer research, as well as writing RESTful APIs to facilitate interaction between terminology-focused graph databases with other internal applications.

Contributed on the front and back end to the development of two key applications within Omni, the Marketing Science and Data software suite used across agencies within Annalect’s parent company Omnicom Media Group. Both applications used Python 3 with Flask for an API driving an AngularJS front-end. Applied Python 3 (including common DS/ML Python packages such as scikit-learn, numpy, Pandas) SQL (Redshift/PSQL), JavaScript (AngularJS), HTML, CSS, Docker, and AWS (Batch, Lambda, Redshift, S3).

Worked on a variety of small and experimental projects within the Annalect R&D ("Labs") team.Created an extension to an existing web application made by the Labs team to produce automated natural language summaries of customer behavior in the consumer decision journey from large sets of demographic and clickstream data, used Keras to drive a recommendation engine type system to explore the relationships between brands' shoppers and browsing behavior.