- Developed and configured deep learning natural language transformers in order to classify categorical and quantitative data into such outputs fields as color, shape, trustworthiness, number, and othersDeveloped speech-to-text data models in order to accurately convert user entered voice recordings into text which would be returned to user facing elements and be fed into other data science pipelines- Communicated the performance of data science models and performed other data analyses in order to to inform product decisions and decision pertaining to data capture- Constructed data science pipeline in order to perform entity recognition with a combination of rules-based and machine learning techniques in order to detect and redact PII fields such as name, address, phone number, email, etc.- Utilized the OpenAI Chat Completions API and engineered prompts using multiple versions of ChatGPT; in order to perform text classifications for fields such as trustworthiness and hostility, and perform summarizations and paraphrasing of text- Determined optimal metrics, analyzed outputs, and synthesized conclusions from A/B testing of UI/UX components for our consumer-facing iOS app and browser using control group testing; ensured statistical integrity by defining clear hypotheses and success metrics, implementing randomization, ensuring sufficient sample sizes, and verifying data quality; contributed to a 10x increase in app downloads and daily active users year over year.- Directed Data Science Ops and MLOps, i.e. efficiently and regularly training models written in Python, utilizing CI/CD and process to rapidly transform the work of data scientists into useable production quality deployments, monitoring model performance and resource usage, and reconfiguring or replacing AI/ML models when superior solutions were discovered

- Conducted research and extracted insights from case studies using our proprietary persona engine using R and Python, in order to inform data scientists of potential new avenues for research- Created advanced unsupervised anomaly detection pipeline that utilized robust feature generation, dimensionality reduction via UMAP, clustering via HDBSCAN, and graph embeddings- Developed supervised machine learning models for use cases including fraud detection, risk modeling, and optimizing the balance between security and user experience; using supervised machine learning techniques including gradient boosting machines, neural networks, support vector machines, and random forests- Conducted research into existing anomaly detection techniques and perform data science experiments in order to inform model selection, feature generation, scalability, model performance, and efficiency- Constructed data engineering and data science pipelines within AWS using services such as EC2s, S3, Lambda functions, SQS, SNS, EKS, and others; and using open-source tools deployed with AWS including Kafka, Airflow, Spark, Docker, Kubernetes, and Helm- Optimized the performance and efficiency of services executed in Python on large high-dimensional data spaces by optimizing parallelization using Dask, using vectorization whenever possible, optimizing data structures, utilizing functions implemented in C when possible, managing memory, re-writing and optimizing some methods within existing packages, and generally utilizing Python best practices- Generated ETLs within Snowflake using SnowSQL and Data Build Tool (dbt) in order to prepare raw data from third-party and proprietary sources into a format optimal to be processed into data science models; operations included joins, aggregations, basic arithmetic operations, statistical operations, data and time calculations, and geospatial operations

- Quantified key mission metrics such as risk, return on investment, optimal areas for future research, probability of successful engagements, etc. by developing regression and classification models using techniques such as neural networks, support vector machines, random forest, gradient boosting machines, clustering techniques and traditional linear and logistic regressions- Performed data profiling and data discovery on client data sets in order to inform client and internal data science teams on data quality, relationships within the data, outliers within the data, significant statistical properties within the data, and to inform the selection of models that could generate valuable insights from the data- Developed anomaly detection in order to detect fraud, risk, insider threats, anomalous user behavior, and anomalous relationships in networks, using traditional statistics, clustering, support vector machines, and neural networks- Architected solutions for over a dozen medium and large federal organizations at all different levels of data maturity, which entailed: designing architectures around legacy systems, hybrid cloud/on-prem architectures, determining to sunset obsolete systems, educating stakeholders on new technology, etc.- Developed over-arching solution architectures with executive clients; led the development of artifacts and implementation plans that incorporate business needs, technology solutions, and data science models- Worked on project to create the first ever unified data warehouse of all VA Call Center data, and developed consolidated analytics dashboards to view comprehensive or drilled-down data insights- Specialized in data analysis and visualization in order to communicate the value of data with interactive and static data visualizations and dashboards utilizing Tableau, Power BI, and R Shiny

• Analyzed complex tactical situations in fast-paced environments• Developed innovative solutions to solve problems in order to meet tactical objectives and timelines • Communicated and coordinated with other government agencies and other global military entities • Supervised the equipment, conduct, and discipline of myself and subordinates• Counseled and mentored employees, and was responsible for their professional as well as personal development• Trained subordinates throughout the unit, and international military personnel, on the employment of various technical systems