Building machine learning models leveraging diverse data sets and APIs to enhance analytics and drive strategic insights for internal and external clients.• Used Generative AI to extract summaries and relevant topics from large numbers of documents by calling GPT3 via an API while providing examples to the model via engineered prompts that ensured the results could be easily processed in Python. • Used Sentence Similarity with Hugging Face language models to identify documents of interest by matching tens of thousands of text fields to label(s) of interest.• Designed Mapbox interactive maps in Power BI by downloading map shape files and data via the census API, geocoding addresses, and wrote them to a MySQL database on AWS.• Implemented Natural Language Processing (NLP) ETL data preprocessing pipeline and models to analyze and summarize large amounts of conversational data and reports using Python and State of the Art Large Language Models.• Wrote Machine Learning extensions in Python for an open-source data analytics, reporting and integration platform.• Created Power BI reports displaying NLP model results for efficient analysis of large numbers of text documents.• Using GPTx Generative AI to create new documents based on related documents which are retrieved using embedding vector search.

• Taught Data Science for Enterprise clients. This involves teaching new content, reviewing student work and providing one-on-one guidance to ensure student success, including enterprise business projects which are being completed as part of the students’ Capstone projects. (Contract position)Successfully completed 12-week Data Science Intensive (DSI) boot camp course. Learned and implemented various machine learning models, including classification and regression models, supervised and non-supervised learning, Natural Language Processing (NLP), decision trees and random forests, and neural nets, among others. • Reinforcement Learning was the most important topic that was not covered in the DSI course. My final capstone project explored this area by creating an algorithm that could teach itself to play a game. A second (brute force) algorithm was implemented to play against the Reinforcement Learning algorithm, which allowed the latter to learn, literally, by its mistakes. The two algorithms (implemented in Python) played 10,000 games against one another, during which the Reinforcement Learning algorithm learned by itself how to win consistently.• Completed client project that used social media to identify areas impacted by natural disasters. This was designed to help agencies like FEMA, working with non-profits, to identify and provide help to people in need. It used Python to retrieve historical geo-tagged social media posts from an API. This covered the time period of an actual hurricane landfall. This data was used to train a Natural Language Processing (NLP) model.

Researched automating KYC Adverse Media Screening using NLP. Designed and implemented enterprise applications using SQL databases and web services.• Reviewed different options to implement Natural Language Processing (NLP) Adverse Media Screening for KYC (Know Your Client). Created proof of concept using Open Source Software to identify risk factors such as money laundering in business news articles, such as from Factiva.• Created generic JavaScript library using TypeScript that encapsulates a NoSQL database to provide local storage of JSON objects and web service connectivity to any HTML UI. This allows a browser-based application to have the performance advantages and responsiveness of a native application, while running as an app on a mobile device such as an iPad. This online/offline library was used by different teams in their iPad apps.• Completed design and implementation of database, web services, and desktop applications for Investment Banking.• Worked with external vendors, such as Microsoft, and various corporate technology teams to resolve serious system issues.

Developed software solutions for insurance applications including Predictive Modeling (Analytics).While working on a Predictive Modeling application, proposed an automated integration test tool and successfully convincing all the stakeholders (management, QA etc) of the value of the project. Implemented the tool, which executed 15,000 tests, ensuring model validity.