Turning fragmented food-system records into a living map of Dallas.
I helped shape the Dallas & Dallas County Food Plan's data and visualization platform—organizing the data foundation, developing geospatial and network views, and translating a complex regional system into a product people can explore.
- My role
- Data architecture, GIS, visualization & project coordination
- Stack
- ArcGIS, Supabase, PostGIS, Cytoscape, Vite
At a glance
The case in under 30 seconds
- Role
- Data architecture, GIS, visualization and project coordination
- Problem
- Food-system records were fragmented across tools, formats and owners.
- Intervention
- Built a shared data model, ingestion workflows, GIS and relationship views.
- Result
- 8,026 organizations and 13,078 locations organized into a living public system.
- Reading time
- 7 min
Evidence note: Counts reflect the current structured platform dataset and implemented map systems.
01 · The context
A systems challenge, not just a mapping challenge.
Dallas County's food system spans public agencies, farms, distributors, food banks, healthcare institutions, community organizations, and residents. Valuable records existed, but they were fragmented across tools and formats.
The work needed to move beyond plotting points. It had to establish a dependable data foundation and reveal both geography and relationships—where assets exist, how actors connect, and where the system may be vulnerable.
Organization, location, and relationship records lived across ArcGIS, spreadsheets, public datasets, and partially manual research.
Inconsistent naming and unclear provenance made the system difficult to update, audit, and explain.
A location map alone could not show how producers, distributors, food banks, institutions, and communities depend on one another.
The platform needed to support present-day exploration while remaining extensible for future engagement and dashboard work.
02 · My contribution
Organize the data. Design the system. Make it legible.
01
Create one coherent data foundation
I organized the system around reusable entities—pillars, characters, organizations, sites, individuals, and interactions—then connected the active working records to a PostgreSQL/PostGIS model in Supabase.
02
Build a repeatable data pipeline
I developed workflows for extraction, normalization, matching, geocoding, validation, and ingestion across ArcGIS, internal workbooks, Dallas Open Data, USDA, and Texas program data.
03
Design complementary visual models
The GIS experience answers where food-system assets are located. The value network map answers how key roles connect, where relationships cluster, and where dependency risks may exist.
04
Turn the system into a public product
I managed the visualization work as a product: modular views, source documentation, dashboard storytelling, stakeholder-ready interfaces, and a structure designed to evolve with the plan.
03 · The platform
One system, three connected layers.
Source layer
ArcGIS, internal workbooks, Dallas Open Data, USDA, Texas Open Data, and stakeholder research
Data layer
Normalization, entity matching, geocoding, provenance catalog, Supabase, PostgreSQL, and PostGIS
Experience layer
ArcGIS GIS map, searchable table, Cytoscape value network, dashboards, and public information pages
Further reading · SMU News
The story behind the maps.
SMU News highlights why Dallas needs a comprehensive food strategy, how human-centered workshops are bringing stakeholders together, and how the value network and GIS maps can reveal connections, gaps, and resilience across the county.
Read the full SMU story ↗04 · The result
A stronger foundation for a living public plan.
The result is not a static visualization. It is a more coherent platform with structured records, documented sources, reusable ingestion workflows, and multiple ways to interpret the county food system.
That foundation makes future updates, stakeholder contributions, analytical dashboards, and new geographic layers more realistic—while giving the team a clearer way to communicate what is known today and what still needs validation.
Key learnings
- 01A civic map is only as trustworthy as the data model and source trail behind it.
- 02Location and relationship views answer different questions; using both produces a more complete systems picture.
- 03Living public infrastructure needs update workflows, not just a polished launch state.
- 04Data visualization becomes useful when it helps stakeholders move from assumptions to specific questions and decisions.
Build for decisions, not decoration
Need to turn complex data into a system people can use?
I connect data architecture, visualization, and product strategy to make complex operations easier to understand and act on.