Full stack developer working with a large, mixed-discipline team of geographers and engineers at USGS-NGTOC. Worked on ETL'd products distributed through The National Map, including 3DEP-derived elevation data and hydrology data from the National Hydrography Dataset. I did my best work in the back-of-the-shop, largely in Python/SQL (GDAL/OGR and the OSGEO stack). I also dabbled in JavaScript (Node/CDK/Angular/Babel/Vue/OGC). I used Terraform (and Gitlab) for DevOps. Our application development was cloud native, much of it on AWS (with some Kubernetes, OpenFAAS, and Apache Spark). We use CI/CD to manage our product delivery and Agile to manage our teams.

I worked with the USFS's FHAAST, GTAC, and EFETAC teams analyzing petabyte-scale remotely sensed imagery; mostly from Sentinel-2, Landsat, and NAIP missions on Google's Cloud Platform (and Earth Engine). In terms of science, much of my work was focused on space-based insect and disease monitoring and aerial surveying. In terms of business, my role was data engineering, focused on building efficient pipelines from resources as various as mylar sheets in an old file cabinet to cloud storage that fed back into our shop's predictive modeling and risk analysis.

Most of my work with PLJV dealt with statistical modelling, remote sensing, and managing inventory and monitoring projects for grassland birds and waterfowl. Because I worked in a region that was almost exclusively privately owned, I also did work with applied resource economics. I used data to find common ground with land owners and to promote healthy economies, sustainable land-use, and good habitat for wildlife.Some specifics of my work included:• Parametric modelling of migratory and breeding-season bird "occurrence"​, occupancy, and abundance using citizen science datasets (e.g., eBird, Breeding Bird Survey) and robust monitoring datasets (e.g., using distance and removal sampling).• Back-end development, spatial modelling, statistics, machine learning, and cartography with 'R'​/Python and C using Earth Engine, GDAL, GEOS, pandas, numpy, and scikit-learn, etc.• Some front-end work with JS• Landscape change modelling focusing on key anthropogenic drivers, including oil-and-gas development, wind energy development, groundwater extraction, and agricultural development.• Mentoring and code reviews for early career analysts, programmers, and interns• Econometric modeling, including short-term supply forecasting, time-series analysis with generalized additive models, and resource depletion modeling with logistic regression.• Led field crews and collected vegetation and wildlife monitoring data to state and federal partner specifications.• Gave regular presentations on research and advocacy to non-technical stakeholders and collaborators at professional meetings.• Grant writing and reviewing for Competitive State Wildlife Grants, Wildlife Conservation Society grants, NRCS CIGs, and ConocoPhilips capacity grants.• Full-stack odds-and-ends: I used Linux (RHEL/Ubuntu), AWS, Docker, GCP, Python, R, Git, PostgreSQL, PostGIS, GeoServer, Flask, QGIS/GRASS, Leaflet, Google Maps API, and Open Layers

My focus at UW was fitting species distribution models to forecast range shifts and suitable climate conditions for three subspecies of big sagebrush. My research also explored the role of climate change on environmental suitability for invasive annual brome grasses and the potential range contraction of suitable habitat for at-risk animal species in the West, such as the Greater Sage-grouse and Gunnison Sage-grouse.Some specifics of my work included:• Regular sampling of sagebrush plant communities throughout Wyoming and Montana. Collected soil-cores for texture analysis in the lab, collected data on vegetation cover; identifying shrubs, forbs, and grasses to species for an ordination study on sagebrush community composition.• Fitting statistical and machine learning models associated with regression, likelihood analysis, and classification tasks. My research involved ensemble modeling with Generalized Linear Models (GLM), Generalized Additive Models (GAM), adaptive regression splines (MARS), maximum entropy (MAXENT), Classification Trees (CT), Random Forests, and boosted regression. I learned that "simpler is better" and that if you can get away with using GLM's or hierarchical (likelihood-based) models, you really should.• Building expert-opinion based demographic models for cheatgrass (Bromus tectorum) for use in spatial models of species abundance given site hydrology and soil factors influencing population growth. • Developed novel methods for pseudo-absence generation, downsampling, and related GIS tasks using the 'R' language.

Responsible for collecting and analysing satellite-derived remote sensing data for central Idaho and helped parameterize and build resource selection (regression) models and survivorship models for radio-telemetried mule deer. Like my role at IB, my work at Idaho Fish and Game involved image analysis and workflow automation -- but the pictures were taken from space and I got the chance to ground truth our remotely sensed imagery using veg data collected in the field. I hung-out with some very bright wildlife ecologists that gave me my first real introduction to population biology and ecosystem modeling.Some specifics of my work included:• Field surveys of vegetation and phenology plots: identifying forbs, shrubs, grasses, and seedlings to species. Collecting rumen samples and site vegetation to characterize ungulate nutrition in the lab. • Geospatial and statistical analysis of mule deer habitat, range, phenology, and survivorship, python-based software development, and geodatabase management using ArcGIS.• Platform testing and software development using virtualization on VirtualBox and VMWare.• Managing GPS collar data using R and ArcGIS. Automated data input tasks by writing custom tools for ArcToolbox in Python.• Designing Python scripts to automatically generate runtime data for the Geospatial Modeling Environment / Hawth's Tools to process large volumes of home range data (automated kernel density estimation). Observations were used to characterize environmental parameters critical to identifying phenology factors in survival models.• Automated raster masking tasks so large volumes of MODIS and AVHRR satellite data describing phenology variables could be processed in hours, where processing by hand used to take weeks.• Automated data management tasks for other department users by creating custom toolboxes using ArcGIS 10's workflow toolkit, Model Builder.

Designed experiments to optimize production efficiency and accuracy, tested experimental reagents and target peptides, and LOTS of microarray imaging analysis for high throughput testing of monoclonal and polyclonal antibodies for folks doing cancer research. I gained a great appreciation for image segmentation work, pattern recognition, and workflow automation using ImageJ, scikit-image, and the Python Imaging Library (Pillow).

Led a team collecting long-term natural seedling regeneration data for sites affected by the Clear Creek Wildfire in the Salmon-Challis National Forest (Idaho). Surveyed plant community composition over 1500 acres using Modified Whittaker Plots and FIA plots. Responsible for teaching field sampling techniques, plant identification, and data entry protocols to technicians. Trained technicians in the use of Trimble GPS units, designed forms for data entry, and taught technicians how to commit field data to geodatabases for reporting to forest service contacts.