Data Scientist

Michele Tobias Portfolio

Geospatial scientist (PhD, Geography, UC Davis) whose Google Sites portfolio documents the crowd-sourced American Viticultural Areas digitizing project she leads at UC Davis DataLab and Library, and splits teaching, code, presentations, and publications into separate tabs.

GISRQGISGeospatial Analysis

Location

Coordinates unknown. Vibes: international.

Credentials

  • PhD, Geography (Biogeography & Geospatial Methods) — UC Davis
  • MS, Terrestrial Resource Ecology & Management — University of Michigan School of Natural Resources
  • BA, Geography (Biogeography) & German (Germanic Linguistics) — UCLA

Skills

GISRQGISGeospatial Analysis

Notable

  • Sautter Award, Silver (2018) for the American Viticultural Areas Digitizing Project
  • Co-author, 'Fourteen things you need to know about collaborating with data scientists', Nature (2023)
  • Repeat FOSS4G speaker, including 'Classifying American Viticultural Areas Based on Environmental Data' (Florence, 2022)
  • UCDavisLibrary/ava dataset repo: 2,104 commits, 81 stars, 59 forks

What makes it work

A breakdown of the choices that make this portfolio stand out.

1

Community/crowd-sourced data project as a leadership signal

The site calls the AVA project long-running and crowd-sourced, built to follow the TTB's verbal boundary descriptions as closely as possible and documented well enough for research use. The linked GitHub repo sits at 2,104 commits and 59 forks. That is distributed data collection turned into a maintained, structured dataset — beyond solo notebook work.

2

Separation of teaching, code, and research

Splitting the site into distinct sections lets a hiring manager jump straight to evidence relevant to the role, e.g. "Code & Projects" for an engineering-heavy screen.

3

Niche technical specialization (geospatial)

Code & Projects is organized by domain — Sandy Beach Vegetation, Geospatial Tools, Wine Geography, Health Geography, Historical Geography — and each project lists its related publications. The same GIS toolkit applied to wine regions, dementia research and 1923 Berkeley fire maps reads as depth that transfers.

What Data Scientists can take from this

Specific, actionable tips to apply to your own portfolio — no generic advice.

  • If you lead or maintain a community/open dataset, document that leadership explicitly — it is a rarer signal than another solo project.

  • Organize a portfolio by function (teaching, code, research) rather than chronology so recruiters can navigate directly to what matters.

  • Own a geospatial or other domain niche if you have one — "GIS-focused data scientist" is more filterable than "data scientist."

  • Workshops and training materials count as portfolio evidence: they prove you can explain technical work to non-experts.

Projects

Every project on the Code & Projects page comes with its paper trail: a Related Publications list under each one, from beach plants to wine regions to the 1923 Berkeley fire. Michele Tobias treats the GitHub repo and the journal citation as two halves of the same deliverable.

Open SourceadvancedMulti-year, ongoing

American Viticultural Areas Digitizing Project

QGISRGeoJSONGitHub

An open-licensed (CC0), community-contributed spatial dataset of every official American Viticultural Area boundary. Each polygon is digitized from the verbal boundary descriptions the Alcohol and Tobacco Tax and Trade Bureau publishes, and the methods are documented so the data is reproducible for research. Contributors submit individual AVA files through GitHub. R scripts scrape eCFR revision histories, check GeoJSON winding order against RFC 7946, merge files, and split them into per-state sets.

This is data-engineering leadership, not a solo notebook: a contribution workflow, a documented method, automated quality checks and versioned releases for a dataset that outside researchers rely on.

  • Follows the legal boundary text as closely as possible. The project site notes that the TTB's own dataset lacks documented methods and retired boundaries.

  • The rcode folder holds the housekeeping scripts: eCFR scraping, winding-order checks, county-field fixes and GeoJSON merging.

  • Output ships three ways: individual AVA files, aggregated files, and per-state files.

TooladvancedMulti-release project (6 releases)

Literature Mapper QGIS Plugin

PythonQGIS 3Zotero APIGeoJSONLeaflet

A QGIS plugin that connects QGIS to a Zotero citation library and writes GeoJSON geometry strings into individual records, so study-site locations travel with the citations. Her Code & Projects page pairs it with a Leaflet web map whose markers and pop-ups load directly from the Zotero Online API, and with a geocoded Zotero bibliography of California beach-vegetation literature.

It shows the full loop: someone who uses GIS tools also wrote one, took it through a peer-reviewed paper, and demoed it with a live map built on the same data.

  • Uses Zotero as the spatial database, so no separate GIS server is needed.

  • The demo map reads marker locations and pop-up text straight from the Zotero API.

  • Presented at FOSS4G 2021 as 'QGIS Plugin Development Is Not Scary: Lessons Learned from Literature Mapper'.

ResearchadvancedConference-paper cycle (2022)

Classifying American Viticultural Areas Based on Environmental Data

RHierarchical clusteringPRISM climate normalsRaster analysis

Tests whether AVA boundaries really reflect the environmental conditions they are supposed to be based on. The AVA polygons are summarized by elevation, precipitation, temperature and soil characteristics, then grouped with hierarchical clustering in R. Inputs include Oregon State's PRISM 30-year normals (precipitation and mean, minimum and maximum temperature) and an 800 m elevation model.

It reuses her own maintained dataset to answer a real policy question with unsupervised methods. That shows a data scientist who builds the data asset and then extracts findings from it.

  • The README shows an example result: AVA boundaries grouped into 6 clusters on climate variables.

  • Built on the AVA Digitizing Project's current boundary file (avas.geojson).

  • Released as a citable repository alongside the paper.

Ready to build your portfolio?

Follow the Data Scientist roadmap — skills, projects, and timeline to get hired.

Data Scientist Roadmap