Gaurav Tuli · PhD Project Portfolio
← Back to portfolio

Global COVID-19 Impact Monitoring

Publication Code Release Policy Media 10X

Scaled the banchmarking and ML for Facebook UMD-CTIS survey to deliver pandemic situational awareness for more than 114 countries.

Role: Principal Data Scientist & Co-first Author

Focus: Behavior Monitoring · Civic Tech · Global Health · ML Classification · Privacy-Aware Analytics · Public Health Intervention · Public-Private Collaboration · Situational Awareness · Survey Design

Outcome: Survey signals adopted by international agencies for COVID-19 situational awareness; co-first authored publications demonstrating methodology and impact.Paper

At a Glance

  • Led Analytics and ML for global expansion of the Facebook UMD-CTIS COVID-19 survey program during peak pandemic response.
  • Benchmarked CTIS signals against public health metrics for 114 countries and built ML models to predict COVID-19 postive tests.
  • Partnered with academia and governments to translate survey intelligence into actionable decisions.

Visual highlights

Global benchmarking of Facebook UMD-CTIS survey signals
Daily survey benchmarking that compared COVID-19 signals across global regions.
Feature pipeline supporting COVID-19 survey analytics
Feature engineering pipeline informing predictive models and data quality checks.

The Problem

  • Official reporting lagged or was inconsistent across countries, obscuring emerging hotspots.
  • Uneven data quality made cross-country comparisons difficult for global health leaders.
  • Policymakers needed privacy-conscious ways to leverage social survey data without compromising respondents.

The Solution

  • Scaled sampling, weighting, and translation operations to reach billions of daily impressions in dozens of languages.
  • Built analytics that benchmarked survey signals against public health metrics to validate usefulness in real time.
  • Implemented stringent privacy, security, and quality controls so partners could act on insights with confidence.

Architecture Overview

  • Established a federated network of survey partners, data pipelines, and governance policies covering 114 countries.
  • Automated ingestion, weighting, and anomaly detection using Python and R, with Bayesian approaches to correct sampling bias.
  • Developed comparison dashboards that aligned survey metrics with epidemiological benchmarks and vaccination data.
  • Embedded privacy safeguards—secure aggregation, suppression thresholds, and QA protocols—before releasing any dataset.
  • Coordinated localization, QA, and partner feedback loops to continuously refine question wording and signal fidelity.

Results and Impacts

  • Provided near real-time pandemic metrics adopted by WHO, CDC, and national agencies for situational awareness.
  • Validated survey indicators against case data, proving policy relevance for mask mandates, reopening, and vaccine planning.
  • Guided resource allocation and messaging campaigns through shared dashboards and rapid partner briefings.

Skills and Tools Used

Technique/Skill Tools/Implementation
Skill/Tool Category Application in Global COVID-19 Impact Monitoring
Global Program Leadership Co-led the Meta/UMD survey consortium, aligning academic, industry, and government partners across 114 countries.
Analytics & Weighting Applied Python, R, and Bayesian weighting to correct sampling bias and track indicators with statistical rigor.
Survey Infrastructure Scaled translation, sampling, and delivery logistics so surveys reached billions of impressions daily.
Privacy & Governance Implemented secure aggregation, suppression thresholds, and QA reviews prior to data sharing.
Stakeholder Engagement Produced daily dashboards and partner briefings that turned survey signals into actionable public health guidance.

Cross-Project Capabilities

  • Demonstrated the ability to operationalize privacy-aware analytics at global scale, informing later federated studies.
  • Strengthened skills in coordinating cross-sector partners—experience reused in vaccine hesitancy and biosurveillance projects.
  • Built reusable pipelines for comparing novel survey signals with official metrics, now applied across other health domains.

Published Papers/Tools

  • Daily dashboards and API feeds powering WHO, CDC, and national COVID-19 situational awareness products.
  • Technical briefs documenting weighting, privacy, and quality controls shared with academic and governmental partners.
  • Co-first authored publications and public reports demonstrating the program’s methodology and policy impact.Paper