Projects.

Attashe

The flagship spatial data provenance platform from Trident Spatial Labs, a spatial data technology company built around trust, provenance, and governance.

Organizations increasingly make consequential decisions on spatial data whose origin, lineage, and reliability are unknown. Attashe addresses that gap: it gives organizations a clear record of where their spatial data comes from, how it has changed, and whether it can be trusted for the decision at hand.

As co-founder and strategy and operations lead, I guide business strategy, partnerships, market positioning, and go-to-market planning, translating complex geospatial and data-governance challenges into clear product priorities.

Community Data Snapshot

An AI agent for place-based nonprofits: enter a service area and a mission focus, and a three-module agentic workflow scopes the most decision-relevant community indicators, researches current values from public sources with live web search, and drafts a funder-ready needs brief.

Responsible AI is an architecture decision here, not a disclaimer. The agent refuses to invent figures, attaches a source and vintage to every value, flags indicators it cannot verify, and requires human review before anything is published.

The pattern is the product: any organization, any geography, the same governed workflow. It is a deliberately small proof of how repeatable, well-governed AI accelerators can serve organizations that will never have data teams of their own.

Global Impact Dashboard 2.0

Co-developed Cadasta’s Global Impact Dashboard, integrating project and CRM data with more than thirty spatial layers into maps, dashboards, and metrics used by partners, policymakers, and practitioners working on land and resource rights.

The automation behind it, built with the ArcGIS Pro Data Interoperability Extension, consolidated previously manual reporting into a single standardized dataset and reduced monthly update time by more than forty hours, freeing a small team to focus on analysis instead of assembly.

Shared-drive governance & remediation tool

Legacy files in cloud shared drives were visible in a browser but inaccessible from Windows: deeply nested folders and long filenames pushed locally reconstructed paths past the 260-character limit. Leadership needed to know how widespread the problem was.

Since those Windows paths only exist when the desktop client reconstructs them, I used AI to rethink the problem and design a script that scanned the shared drives, traced each file through its parent folders, rebuilt the effective Windows path, and flagged violations and near-misses, producing a quantified evidence base across thousands of files. The tool then translated the organization’s naming conventions into remediation rules and generated compliant renames, flagging conflicts and presenting options instead of deciding autonomously, so every recommendation stayed under human review.

The audit revealed the real issue underneath: inconsistent structures, uncontrolled filenames, and institutional knowledge living in people instead of systems. The outcome was a human-guided remediation system and a quantified view of governance debt that leadership could act on.

The AI Impact Wire

A curated editorial portal tracking AI for good: plain-language briefings, research, funding opportunities, jobs, events, and practical use cases across climate, health, education, equity, governance, and data ethics.

The portal monitors trusted public-interest sources including the UN, OECD.AI, WHO, UNESCO, Stanford HAI, and Partnership on AI, so mission-driven teams can quickly see what is new, who published it, and why it matters. AI-assisted summaries are reviewed by a human editor before publication, with a public editorial and AI disclosure: the same responsible-AI principles that run through the rest of this work, applied to the publication itself.

AI and data for famine early warning

Data management leadership on the FEWS NET 8 Decision Support Project, the U.S. Department of State funded food security early warning program that informs humanitarian response decisions worldwide.

The work spanned data governance, GIS and ICT strategy, and stewardship of the FEWS NET Data Warehouse: a semi-automated audit of unpublished datasets that surfaced governance bottlenecks and orphaned data, rationalization of ArcGIS Online governance across legacy accounts and access risks, unit-of-analysis governance across five countries, and migration of legacy network-drive GIS workflows to cloud infrastructure, supporting a team that grew from three to eight specialists.

GDHI / WRSI workflow modernization

The Global Dashboard for Hotspot Identification (GDHI) used Water Requirement Satisfaction Index (WRSI) data to support crop-production analysis across Ethiopia, Kenya, Somalia, and Uganda, but it ran on brittle batch files, scattered Python scripts, hard-coded years and product settings, and errors that failed silently midway through a run.

I used generative AI as a development and diagnostic partner to reconstruct the workflow, surface embedded assumptions, and redesign the architecture, while I remained responsible for the methodology, requirements, validation, and final behavior. The rebuilt system runs from a single orchestrator with a configuration file and consolidated scripts: a user enters a date and the tool selects the right products, verifies availability on the source server, validates the geodatabase before each run, and captures everything in a timestamped log with human-readable errors.

The result moved risk to the start of the process, where it can be caught before analysis runs. AI accelerated the code analysis, redesign, debugging, and documentation. Domain expertise and human validation stayed responsible for the rules and the results.

OASIS

The Operational Activity Security Information System: a U.S. Department of State funded security information platform used by United Nations agencies and NGOs operating in Iraq and Afghanistan.

Conceived at the Vietnam Veterans of America Foundation and proven through a two-year proof of concept that won federal funding, OASIS gave humanitarian organizations a shared, internet-enabled picture of operational security: hazardous areas, minefields, attacks on aid workers, and route planning, in a form usable by non-technical field staff. It was ultimately used by organizations including UN OCHA, UNDP, UNICEF, CARE, World Vision, and the Red Cross across Afghanistan, Colombia, Iraq, Georgia, and Pakistan, and was named a World Bank Development Gateway Award finalist in 2005, one of five selected from more than 150 nominees.

Two decades before responsible AI entered the vocabulary, OASIS demonstrated the same thesis this practice is built on: shared, governed information infrastructure changes what organizations can safely do.