Selected work

From complex data to decisions people can defend.

A curated portfolio spanning development economics, healthcare financing, experimentation, pricing, and applied machine learning.

Each project is framed around the decision, the evidence, and the outcome—not just the algorithm.

Featured case study

01
Illustrative food-crisis location verification queue

Development economics · Humanitarian early warning

Food-crisis early warning from bilingual news

Designed an auditable screening system that converts English and Arabic news into location-level evidence for food-security analysts across Iraq, Jordan, Lebanon, Palestine, and Syria.

  • Validated and translated 167 risk factors across 12 themes.
  • Resolved mentions against 357 geographic entities.
  • Separated signal strength from evidence confidence.
  • Connected calibrated risk to policy-facing action thresholds.
Read the case study

Applied impact

02

Healthcare AI · Rwanda

Claims-driven health financing and provider performance

Built decision-support systems using claims and provider data to support financing reform, utilization monitoring, prescribing quality, and operational oversight.

  • 9M+insured people represented
  • 100M+claims processed
  • 1,000+providers monitored
Health financingClaims analyticsAnomaly detectionDecision support

Commercial ML · Europe

Pricing leakage detection and optimization

Developed anomaly-detection, elasticity, uplift, and optimization models for commercial decisions across a large industrial product portfolio.

  • €11M+annual leakage protected
  • €500M+portfolio modeled
  • 13%gross-profit improvement supported
PricingUplift modelingPySparkDatabricks

Digital health · Emerging markets

Engagement, forecasting, and adaptive interventions

Built forecasting, survival, and experimentation workflows for pharmacy supply chains, healthcare workers, and mobile-health applications.

  • 18%reduction in stockouts supported
  • KDD ’22engagement research published
  • Adaptivebandit and intervention designs
ForecastingSurvival analysisA/B testingBandits

Engineering projects

03
Recommendation-system architecture diagram

Multimodal AI · Recommenders

Multi-Modal LLM Product Recommender

Combined review text, metadata, images, and temporal signals for next-item prediction across approximately 3M interactions.

View project
Abstract visualization representing adaptive decisions

Experimentation · Reinforcement learning

Deep Contextual Bandits

Reusable experimentation workflows built around Bayesian neural networks, Thompson sampling, and decisions under uncertainty.

View project
Customer segmentation visualization

Customer analytics

Segmentation and retention intelligence

Cohort, attribution, RFM, churn, and satisfaction analyses translated into practical commercial recommendations.

Text-analysis visualization

NLP · Research

Text as evidence

Applied text retrieval, classification, similarity, and network analysis to public-policy and customer-behavior questions.

Selected research

Published and presented work

Need the detailed chronology?

See the experience, methods, and tools behind the work.