Humanitarian Aid Data Collector with Mastra AI and GPT-5 Pro
Addressing aid shortfalls highlighted in The Lancet Digital Health research, this project requires building an ethical AI data collection engine for conflict and disaster zones. Gathering ground-truth field data in humanitarian contexts requires stringent privacy preservation, bias detection, and differential privacy guarantees. Developers will utilize Mastra AI with GPT-5 Pro models to build workflows that sanitize sensitive demographic inputs, verify source authenticity, and synthesize equitable resource distribution recommendations.
What you are building
The core problem, expected build, and operating context for this challenge.
Construct an ethical, privacy-preserving humanitarian data pipeline using Mastra AI and GPT-5 Pro to optimize aid allocation.
How work is evaluated
Evaluates privacy retention, field report verification accuracy, and resource distribution fairness.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
pii_leakage_check
Ensure zero personally identifiable information remains in output
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
utility_preservation
Ratio of statistical utility preserved post-anonymization • target: 0.9 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Implement privacy-preserving data extraction workflows using Mastra AI and TypeScript
Integrate GPT-5 Pro reasoning to detect biases in field report datasets
Enforce differential privacy and anonymization protocols on PII data streams
Generate equitable aid allocation matrices under constrained operational budgets
Reference links and supporting material
Synthetic crisis zone field surveys, village population registers, and disaster response logistics datasets.
Challenge setup
Evaluates privacy retention, field report verification accuracy, and resource distribution fairness.
Challenge input
JSON array of raw unstructured field survey entries
Mastra AI
Mandatory TypeScript framework
GPT-5 Pro
Assigned high-tier reasoning engine
Evaluated output
JSON object with anonymized_records and fairness_score
- Ensure zero personally identifiable information remains in output
- Ratio of statistical utility preserved post-anonymization • target: 0.9 • range: 0-1
- Benchmark: EthicalAI-Humanitarian
- Utility Preservation target: 0.9
- 1 public reference case
- JavaScript execution harness
- javascript sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- GPT-5 Pro · Required
- Mastra AI · Optional
- GPT-5 · Optional
- Mastra AI · Optional
Evaluation contract
- pii_leakage_check · Weight 1
- utility_preservation · Weight 1
Metrics
- utility_preservation · Target 0.9
Benchmarks
- EthicalAI-Humanitarian
- security
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON array of raw unstructured field survey entries
Output: JSON object with anonymized_records and fairness_score
Execution harness: JavaScript
Evaluation tools
- Mastra AI · Agent Framework
- GPT-5 Pro · Foundation Model
Recipe freeze marker
No recipe freeze time is recorded.