• Work with USSOCOM J8 Manpower and Strategic Planning on building a machine learning document summarizer. Built four different prototype models that successfully extract text from PDFs. • Researched and built a text corpus to train and test the text summarizer model.• Created a data visualization tool for the 5 years Program Objective Memorandum (POM) in Advana/Qlik and Power BI. • Act as liaison between J8 personnel and the CDAO Data Maturity Assessments Team.• Catalogue Special Operation Forces (SOF) databases, into the Department of Defense (DoD) federated data catalogue, to identify DoD authoritative data sources.• Explain abstract concepts, about data science, that are understandable to non-data science people.• Ensure that SOF data sources are VAULTIS compliant:1. Data is Visible – Consumers can locate the needed data. 2. Data is Accessible – Consumers can retrieve the data. 3. Data Understandable – Consumers can recognize the content, context, and applicability. 4. Data is Linked – Consumers can exploit data elements through innate relationships. 5. Data is Trustworthy – Consumers can be confident in all aspects of data for decision-making. 6. Data is Interoperable – Consumers have a common representation/ comprehension of data. 7. Data is Secure – Consumers know that data is protected from unauthorized use/manipulation.

• Investigate and discover patterns and correlations within the Department of Transportation’s Federal Railroad Administration (FRA) accident database. Factors that drive accidents are used to build predictive models.• Perform dimensionality reduction techniques on the FRA database. Apply principal component analysis to select variables that can be deleted in order to work with a cleaner dataset.• Build a geospatial dashboard, using Dash, to display the intensity of railroad disasters in relation to locations of Department of Defense installations and critical infrastructure.• Build natural language processing topic modelers and detectors using data from open sources such as Wikipedia, Miller Project, Associated Press, Twitter, ABC News and Project Gutenberg.• Collect scientific articles, with appending abstracts, from the PLOS.org, NIPS and arxvis.org research repositories. This is to identify trending subtopics by running the articles through unsupervised topic modelers. The methods used are Non-negative Matrix Factorization (NMF), Term Frequency–Inverse Document Frequency (TF-IDF), Bidirectional Encoder Representations from Transformers (BERT), Latent Semantic Indexing (LSI), Latent Dirichlet Allocation (LDA), Singular Value Decomposition (SVD), and Probabilistic Latent Semantic Analysis (pLSA).• Create and deploy functions to measure the coherence scores for the topic models. The coherence scores measure the interpretability of the topics and allow the end user to know when model parameters have been optimized.• Build a supervised fake news detector using news articles verified and tagged as being real or fake.• Build a topic modeler using the t-distributed neighbor embedding (t-SNE) visual to show how topics, of the United States presidential speeches, change over time from George Washington to Joe Biden.

• Develop predictive models for military operation applications. • Guide team on understanding what data skills are important in supporting machine learning (ML)/ artificial intelligence (AI) models. • Demonstrate ML/AI models for the Command stakeholders. • Explain the technicalities and purposes of ML/AI projects to both technical and non-technical personnel. • Conduct staff orientation training about Data Modernization to new hires. • Acquire and expand virtual staging and production servers to manage big data. • Write memorandums of agreement and interconnection security agreements to build application user interfaces (API) with government partners for secure data exchanges. • Store large datasets in clous environments such as Microsoft Azure, Snowflake and Apache Spark. • Build text classification and topic modeler using natural language processing (NLP). • Gain knowledge of tools such as Scikit Learn, Pytorch, TensorFlow, Bert, and Spacy. • Coordinate the Data Council of Colonels and the Data Board of Directors for final decisions regarding data. • Manage and update the Commanders Data Board dashboard with key tasks, milestones and meeting minutes using C2IE (Command Control Information Environment). • Set up and deploy ML/AI models in AWS, Jupyter Notebook, Project Maven, and DataRobot for experimentation. • Use Microsoft Project and Teams to break ML/AI projects into subtasks and assign to fellow team members. • Employ techniques and theories drawn from mathematics, statistics, from the subdomains of machine learning, classification, cluster analysis, and data mining.

• Provided geospatial analysis and statistical expertise within the USSOCOM J-39 Mission Information Support Operations (MISO) branch to support MISO/GEOINT programs and objectives, assist with program assessments, and ensure scientific rigor. • Conducted data mining tasks of preprocessing and dimension reduction. • Utilized data mining tools of single vector models (SVM), K-Nearest Neighbor (KNN), Decision Trees & Rules, Artificial Neural Networks (ANN), Ensemble Learning, Cluster Analysis, Association Rules, Web Mining and Text Mining. • Built machine learning algorithms through a combination Python, R, SPSS and of ArcGIS Geostatistical Analyst to make meaning of spatial and non-spatial data. • Utilized data science testing methods to evaluate which machine learning algorithms provide the greatest predictive powers. • Delivered geostatistical expertise on analysis of variance (ANOVA), scatterplots, stem and leaf outlier testing, A/B testing and bivariate correlation calculations. • Performed advanced statistical analysis in commercial and open-source environments such as Python, SPSS, R and SAS. Utilize ArcGIS extensions (e.g., Spatial Analyst, 3D Analyst and Network Analyst) to make inferential conclusions on events and their relations to population demographics, terrain, and country level economics. • Used Python to create Twitter APIs and web scrappers that can pull in website content into tabular formats that can be used as training data for supervised multi-text classification machine learning models. • Conducted machine learning experiments, in text classification, on focus group transcripts, using SciKit Learn, NumPy, SciPy and Matpotlib Python modules. • Built a machine learning conflict prediction model, of the Syria AOR, which test the performance of random forest and naïve Bayes learning algorithms. Created written and oral presentations, of data findings to end users, in a professional and easily understandable format.

• Planned, coordinated, and executed the project management duties for Afghan National Army (ANA) and Afghan National Police (ANP) service and construction projects and contracts throughout 14 provinces in Eastern Afghanistan.• Advised and mentored 25 engineers and 3 engineering teams needing funds for construction and services in order to maintain and operate ANA and ANP training facilities.• Provided oversight of over 93 permanent and 120 planned construction projects valued at $1.964B.• Backfilled the duties of the Chief Engineer by advising higher leadership on over 110 infrastructure projects valued at $68M. • Executed a formal “make or buy” analysis for ArcGIS 10 mapping software and followed the procurement processes from planning to closing.

• Prepare long and short-term enrollment projections to drive decisions in operations, capital budget and rezoning efforts• Manage district’s enrollment, housing, demographic and attendance boundary data• Regularly employ validation techniques to ensure data quality and integrity • Create classroom size and budget options in order to facilitate discussions on staffing• Perform statistical analyses using statistics software (e.g., SPSS, SAS)• Conduct research on variables that correlate to enrollment projection accuracy• Test accuracy of various enrollment projection methodologies• Disaggregate and query 2010 Census data, through MS Access, for research projects• Communicate to APS stakeholders in a clear and concise manner about small area population dynamics and its effects on short and long-term construction projects• Ensure schools conduct greater accountability when entering student demographic and housing information into district’s database• Research and advise executive leadership on policies dealing with student enrollment, eligibility and privacy• Assist transportation department in school bus routing and selecting optimal bus stop locations based on demand• Expert user and training mentor of ArcGIS 10 (Geographic Information Systems) mapping software used to professionally and accurately delineate school attendance boundaries and illustrate how student households are spatially located throughout the district• Subject matter expert on hiring committee to recommend GIS Manger• Lead selection committee when district commissions studies from outside agencies • Serve as project manager over various studies to ensure they stay within scope, schedule, cost and quality

• Performed quantitative consumer market research by segmenting the north Georgia market in twelve different target consumer groups.• Designed, planned, executed, and analyzed geographic and consumer data to ensure land acquisition specialists receive the greatest return on investment.• Catalogued existing and future housing development into ArcGIS and Map Info and provided leadership with weekly reports, which illustrated the health of the north Georgia housing market.• Provided what-if scenarios of housing subdivision locations against aerial photos in order to determine market cannibalization.

GIS Technician - Atlanta Gas Light (contract), Atlanta, GA October 2002 – May 2003• Transferred as-built conduit drawings into a collective GIS database by converting drawings into TIFF files, registering the images and drawing the lines with desktop GIS tools. • Updated and verified various land and conduit attribute information in order to build a searchable utility database.