Nathan Dzbenski Email & Phone Number

Developing cutting edge physics for muon tomography using GEANT4 and Garfield++ as well as machine learning imaging techniques for cleaning everything up. Like Superman’s x-ray vision if he ate a lot of carrots.

Greensboro, North Carolina, United States

• Develop cloud-first (GCP, Azure, AWS - yes, we use all three) machine-learning (ML) pipelines (from raw data to publication/dashboards) trained and tested on big data utilizing both AGILE and DevOps development practices in order to deliver accurate, trustworthy predictive models from which stakeholders can make very important decisions • Ongoing research includes predicting student churn using both traditional and multi-layer perceptron, or MLP, artificial neural-network (ANN) for the next semesters in order to produce lists for outreach in order to increase retention. Case studies on ethics in these models (i.e. their probability of producing results that perpetuate established cultural biases due to inclusion of some demographic information) are done in parallel to attempt to minimize bias in the models.• Utilize Terraform (or Docker), GitHub (or BitBucket), Jenkins, as well as GSuite, Box, and Microsoft Teams all in order to develop CI/CD analysis code that ensures ease of collaboration between small and large groups within the University as well as feedback loops that result in improvements• Long-Short Term Memory (LSTM) and other Recurrent Neural Network (RNN) time-series forecasting frameworks have been written to predict Virtual Machine performance/failure within IT Services Infrastructure• Sentiment analysis code was produced for surveys, feedback, IT service-ticket requests, and other text-based responses via Natural-Language Processing (NLP) methods using both “traditional” and neural-network approaches• Develop ML feature-engineering and dimensionality-reduction frameworks that ultimately lead to tree-based predictive models of student behaviors such as retention probability and graduation rateDevelop ML outlier and anomaly-detection/extraction algorithms through k-Nearest Neighbor (kNN) and other clustering methods

• Developed Monte-Carlo simulation software for high-energy physics particle detectors using C++, Perl and XML while collaborating with others using Git repositories• Accessed and modified databases of physics calibration constants with Python and MySQL• Utilized finite-element method (Elmer and ElmerSolver) and numerical methods (4th-order Runge-Kutta) to solve the electrostatics coupled-partial differential equations on the nodes of geometric meshes• Developed modeling software using C++ and utilized COMSOL, Ansys, and GMSH to understand the characteristics of particle physics detectors• Engineered and constructed electron drift chambers and electronics using Garfield++, SketchUp, AutoCAD, AutoDesk Eagle (for circuit design)• Data analysis of a large data sets to optimize detector efficiency, extract predictive models of physics behavior using regression analysis and complex function fitting (Gaussian, Landau, etc) with C++, Java, Python, MySQL, shell scripts, and Auger jobs• Utilized neural networks and deep learning to assist in detector optimization and for particle track identification• Implemented Helix fitter and Kalman Filter for track fitting using C++ in order to extract particle momenta in a variable magnetic field• Developed webpages for distribution of data analysis and research using JavaScript, CSS and HTML

Norfolk, Virginia Area

Oversee, instruct, and guide introductory physics students through the laboratory portion of this course.

Newport News, Va

Worked with the GlueX collaboration in experimental Hall D at JLab assisting in the construction and testing of the low-granularity pair spectrometer.

Combat deployments to Middle East and Central Asia.

Doctor Of Philosophy - Phd, Experimental High-Energy Physics

Master Of Science - Ms, Physics

Bachelor Of Science - Bs, Physics