Greg Werner Email & Phone Number

Severn, Md, Us

Severn, Md, Us

Fairfax, Va, Us

Washington, District Of Columbia, Us

Worked on contracts with Veterans Affairs (VA) and the Social Security Administration (SSA). Worked at a Leidos facility with Leidos employees for my SSA work . Came up with and researched new Natural Language Processing (NLP) problems for each client to bring value to their mission. Was responsible for creating new problems based on customer needs and then seeing them through the entire data lifecycle up to development and proposal of solutions. Programmed mainly in Python with Jupyter and Eclipse using conventional machine learning libraries including pandas, spacy, scikit-learn, keras, tensorflow. Used both established supervised algorithms (i.e. Logistic Regression, Decision Trees, Gradient Boosting) as well as deep learning techniques (CNN, RNN, LSTM). Worked on many BERT derived libraries/Git repositories, including BioBERT which was used to handle medical data in disability determinations. Also investigated OCR of handwritten documentation to support data collection, especially in the case of handwritten doctor notes.

Worked with custom natural language understanding (NLU) technology based upon Google DialogFlow to detect intents and entities for an in-house chatbot. Refined training utterances (quanitiy and linguistic composition) to train custom NLU models. Was part of surge that helped roll out verson 1.0 to all the clients in June which was great success. Researched alternatives for custom NLP piece and led team of 3 NLP engineers for version 1.5 of product which included work with Rasa, spaCy, ElasticSearch, Stochastic Gradient Descent (SGD) for online learning and Microsoft Luis. Coded in Python, with some Java as necessary.

Tackled mainly NLP specific problems encountered by company. Worked on the topics of named entity disambiguation, custom named entity recognition, sentiment analysis at the document, sentence and entity levels and corporate/financial binary classifiers. Created in-house stacking algorithm for document level sentiment analysis. Contributed to the field of competitive intelligence. Programmed daily in C#, Java and Python. Used Stanford CoreNLP extensively. Created manual annotations. With a fellow co-worker, co-authored a powerful general-purpose annotation tool, Agent Annotator, to help with semi-automated document analysis and annotation which is on my GitHub site.

Annapolis, Md, Us

Improved custom pipeline based on Stanford NLP. Worked with cross-document correlation. Assisted in submitting proposals. Helped develop custom ontology. Explored implicit causal relations between entities. Looked at ways to modernize NLP stack including in-depth research and prototyping into spaCy. Helped develop annotation guidelines for named entity, relation and event annotation. Designed a few regex based annotators, including most notably a phone number annotator which can represent up to 20 countries simultaneously. Performed work primarily in Java, but also did prototyping in Python and R.

Developed Java 1.8 client/server application using RESTful web services for Search and Rescue purposes. Apply Geographic Information Systems (GIS) theory in performing coding duties. Specializing in UI design / front-end of the application, using JavaFX 8. Work in an agile software development environment

Washington, D.C., Us

As a PhD candidate, worked on Deep Exploration and Filtering of Text (DEFT), a DARPA project. Researched level of committed belief (LCB). Applied work to analyze and detect medical patients who might be suffering from schizophrenia. Performed development for a Java desktop application. Used maven and git on a Linux platform with Java 1.7. System had very intense memory requirements and involved large amounts of data. Worked closely with other PhD candidates and professors. Did some prototyping in MATLAB and R.

Reston, Virginia, Us

Lent Natural Language Processing (NLP) support/expertise to customer for in-house project. Design, develop and lead coding on web applications using Java, PHP and C#. Provide desktop application support using a wide variety of programming languages including: C#, Python, C++, PHP, Perl and Java as needed. Led a team of 4 developers. Help define requirements and specifications for projects.

Abd, Computer Science

Ms, Computer Science

Bs, Ee, Computer Science, Math