Bast builds explainable, accurate Conversational AI Technology (CATs) with its unique AI Engine.Led by the brilliant Beth Rudden, Bast AI is building an ethos and infrastructure platform of trustable AI to augment, not replace, humans. What would you build with AI if you could?

Water in Grand Canyon travels through complex underground cave and karst systems into aquifers, springs and seeps. Understanding these pathways supports forecasting the park's water supply, predicting recharge areas to underground aquifers, and protecting park water resources.I worked closely with park hydrology, climate science and GIS domain experts to craft machine learning explorations of a recently organized geodatabase of data collected at hundreds of hydrology sites throughout the park. Piping data from this Esri geodatabase into Python pandas, cluster analysis and random forest machine learning algorithms revealed patterns in geochemistry suggesting links between specific hydrology sites. These similarities provide insight to inform best use of the park's testing resources and dye trace studies to confirm expected water pathways. We also explored geochemistry to potentially predict uranium presence at sites where uranium had not yet been measured and found some expected geochemical correlations. Challenges included sparse data and wide variation in variables measured at different locations. This was a short term project intended to detect patterns in the dataset to better inform park domain experts planning park hydrology research and field work. National Park Service Research Permit # GRCA-2024-SCI-0054

I took a sabbatical after my mother passed away following a long illness and began retooling my skills for Data Engineering, MLOps, and AI infrastructure.I began building an automated, containerized data solution with a microservices architecture using Docker, Python, and SQL for loading NOAA weather data into PostgreSQL and exploring how to build infrastructure for transparent and explainable AI.

• River Sediment Data Analysis: Created data visualizations and performed statistical analysis on river sediment data from the US Forest Service using Python pandas, matplotlib, and Jupyter.• Predicting Dementia: Built machine learning models to predict dementia from spoken text with Python pandas, sklearn and Jupyter.

Master of Information and Data Science• Predicting Solar Panel Adoption: Developed Python random forest model with hyperparameter tuning and extensive feature engineering to improve predictions for US residential solar panel adoption, scaled to distributed architecture using Spark, piped to Tableau via SQL as a geospatial tool.• Logistic Regression with Gradient Descent in Spark: Wrote and deployed MapReduce gradient descent machine learning algorithm at scale on distributed cloud infrastructure using logistic regression to predict online ad click-through rate. Algorithm and feature engineering written using Spark and PySpark in Jupyter Notebook. Cloud computing with Google DataProc.• Streaming Twitter Sentiment Analyzer: Designed and implemented machine learning pipeline for real-time Twitter sentiment analysis, using IBM Cloud, Kafka, Spark Streaming, Cassandra, StanfordCoreNLP, and Tableau.• Big Data Architecture - Exploring Solar Energy Production: Built ETL pipeline to explore weather impact on solar energy production using AWS, PostgreSQL, Linux, and Python packages psycopg2, pandas, and sklearn.• Data Ethics - What Can Data “Know?": Wrote white paper exploring multiple interpretations emerging from seemingly straightforward sexual assault crime statistics, inspired by Sandra Harding’s concept that “All scientific knowledge is always, in every respect, socially situated.”

• Built a Night Skies science program in collaboration with Grand Canyon National Park's Science and Resource Management division to protect night sky visibility and nocturnal wildlife from light pollution. Work featured in Arizona's largest newspaper, The Arizona Republic.• Designed data strategy, data architecture, built geodatabase, and collaborated on data analysis and modeling using ArcGIS and Python for a park-wide assessment of exterior lighting, a critical component of the park’s subsequent Dark Sky Park status.• Database planning included frequent collaboration with multiple stakeholders, including park architects, electricians, project managers, backcountry rangers, law enforcement rangers, and resource protection experts to ensure that data collection and analysis would satisfy needs and expectations of all stakeholders and end users.• This database became the foundation for a cooperative agreement between the National Park Service (NPS) and the International Dark Sky Association (IDA) to create lighting assessment protocols for use throughout national and state parks. • Supervised volunteers collecting data in remote wilderness locations.• Co-authored Night Skies Visibility Modeling with Grand Canyon's GIS Program Manager, accepted for presentation at the 2015 ESRI User Conference.• Co-authored Protecting Natural Lightscapes with NPS National Night Skies Team, presented at the 2015 George Wright Society Conference on Parks and Protected Areas.• Created and presented a summary of Grand Canyon’s Night Skies Protection program for the Grand Canyon Field Institute’s 2015 annual Hiking Guides Training Seminar.

• Photographed musicians and personalities for media and community organizations. Clients included The Museum of the City of New York, Macy's, Queens Council on the Arts, The Tate Group, Bedford Stuyvesant Restoration Corporation, People, Relix and Global Rhythm magazines.• At Time Inc. Digital Photo Studio, maintained a digital image database with hundreds of thousands of images, including quality control over archival procedures, color correction, and image metadata.• Designed data visualizations and marketing materials using PowerPoint and Adobe InDesign for advertising agencies, investment banks and law firms, including Havas Creative, Ropes and Gray, and D.E. Shaw & Co.• Taught Photoshop and wrote curriculum as an Adobe Certified Photoshop Instructor at Noble Desktop.