Data Saves Lives (DSL)

Data/Map
Data/Map
Data/Map
Sentinel Chat Agent
Landing Page
Data Chart
Data Chart
Data Chart

DSL — Bridging the Humanitarian Funding Gap Elevator Pitch A decision-intelligence platform that identifies "forgotten" humanitarian crises by mapping the structural mismatch between crisis severity and donor funding, enabling UN advocates to generate low-bandwidth "Crisis Alerts" in seconds.

Inspiration

When we spoke with Mary (UN representative), she pointed out a heartbreaking reality: crises aren't just overlooked because of a lack of money, but because of a mismatch in data. A country could have a high severity score but zero "funding velocity." We were inspired to move beyond static charts and build a tool that actually "flags" these neglect-zones automatically.

What it does

GeoInsight connects OCHA’s Financial Tracking Service (FTS) and INFORM Severity data via a Databricks Genie AI Space. It allows users to ask natural language questions and receive:The "Red Flag" Analysis: Automatically identifies countries with high severity ($>4.0$) and low funding ($<25\%$).

The Structural Gap

Calculates the difference between "People in Need" and "People Targeted." Benchmarking Toggles: Compares crisis types (e.g., Flood vs. Flood) across different regions to ensure funding asks are reasonable.Low-Bandwidth Crisis Alerts: Generates lightweight Markdown/Text summaries for field officers working in areas with poor internet connectivity.

How we built it

Data Lakehouse: Hosted on Databricks, utilizing Unity Catalog to manage humanitarian datasets.Intelligence Layer: Built a Databricks Genie Space with custom SQL Measures to calculate the "Neglect Index" and "Funding Velocity."Frontend: A real-time website communicating with the Databricks API via a secure backend.

The Logic##

Implemented Decision Intelligence by defining specific quadrants in scatter plots to visualize "mismatch" metrics.

Challenges we ran into

The "Owner" Access Paradox: We faced a major hurdle with Databricks authentication and registration IDs (c74296c7...), which we had to troubleshoot by re-syncing our Azure Entra ID permissions.Visualizing the Invisible: It was difficult to make "low funding" look as urgent as "high severity." We solved this by creating a custom Neglect Index formula.API Connectivity: Overcoming the "Unexpected token <" error by ensuring our SQL Warehouse was serverless and always ready for real-time website requests.

Accomplishments that we're proud of

Successfully translating Mary's "dream metrics" (like Funding Velocity) into functional SQL logic. Creating an exportable "Crisis Alert" that respects the bandwidth constraints of field workers.Hitting the 100% Benchmark on our Genie Space instructions to ensure the AI never "hallucinates" humanitarian stats.

What we learned

We learned that in the humanitarian sector, clarity is more important than complexity. A simple Markdown table can save more lives than a 50-page PDF if it reaches the right donor at the right time.

What's next for DSL

Media Sentiment Integration: Linking GDELT media mention frequency against funding levels to identify "Truly Forgotten" crises.Predictive Alerting: Using machine learning to predict when a "Funding Stall" is about to happen before the crisis worsens.