I turn raw spatial data into decisions people can act on.
Geospatial analyst working across QGIS, PostGIS and Python. I build reproducible notebooks for land cover classification, hydrology, and urban growth modeling — then package the findings so planners and non-technical stakeholders can actually use them. Most weekends I'm out testing the terrain myself, on foot or by kayak.
Selected work
A case study, plus three recent projects spanning remote sensing, hydrology, and municipal planning.
Predicting modern locations from a 1930s plat map, using OCR and a street buffer
A scanned plat has no coordinates and handwriting that runs in every direction. This walks through using an ordinary GIS buffer to narrow an OCR problem, then matching what's left against the modern street network.
read the case studyPython notebooks
Documented and reproducible. Clone any of these and point them at your own AOI or dataset.
import rioxarray as rxr
# load and clip Sentinel-2 stack to AOI
aoi = gpd.read_file("aoi.geojson")
stack = rxr.open_rasterio("s2_stack.tif")
clipped = stack.rio.clip(aoi.geometry)
ndvi = (clipped.nir - clipped.red) / (clipped.nir + clipped.red)
ndvi.plot(cmap="RdYlGn")
NDVI Time Series
Five-year vegetation trend across a reservoir watershed
# features: slope, distance to floodplain, soil class
X = parcels[["slope", "fld_dist", "soil_k"]]
y = parcels["flood_flag"]
clf = RandomForestClassifier(n_estimators=300)
clf.fit(X, y)
parcels["risk_score"] = clf.predict_proba(X)[:,1]
Parcel Flood Risk Model
Classifier scoring 40k parcels for a resilience grant application
import networkx as nx
# pull walk network, build 15-min isochrones
G = ox.graph_from_place("Eau Claire, WI", network_type="walk")
centers = gpd.read_file("clinic_sites.geojson")
for pt in centers.geometry:
node = ox.distance.nearest_nodes(G, pt.x, pt.y)
subgraph = nx.ego_graph(G, node, radius=900)
Clinic Access Isochrones
15-minute walk-access sheds around rural clinic sites
ee.Initialize()
# composite + classify a Sentinel-2 scene in GEE
img = (ee.ImageCollection("COPERNICUS/S2_SR")
.filterBounds(aoi).filterDate("2025-06-01", "2025-09-01")
.median())
classified = img.classify(trained_clf)
Land Use / Land Cover Classifier
Random forest classification over a growing-season composite
About
Background, tools, and how I like to work.
I'm a recent graduate of the University of Minnesota, with a B.S. in Geography (GIS) and Plant Science. I focus on the Python side of GIS — automating workflows to streamline tasks that would otherwise mean a lot of manual digitizing.
My background spans both plant science and GIS, which gives me a toolset for tackling large-scale spatial problems with real scientific grounding behind them — not just the geometry, but why it matters ecologically.
Outside of work I'm usually spending time enjoying what sparked my interest in this field in the first place — the great outdoors. Hiking, kayaking, and plant identification along the way make me feel more in tune with the landscape, and with myself.