1. **GMV integrity fix (Noon):** Independently caught a 4x GMV inflation bug running across three production pipelines, which protected reporting accuracy for category leadership.
2. **Reactivation revenue (Noon):** Led the analytics behind a customer reactivation initiative that preserved ~₹100M in revenue.
3. **Cross-sell cohort (Noon):** Built a ~190,000-customer cross-sell cohort for targeted campaigns.
4. **Inventory planning (Noon):** Found that the top 20 brands drive 80%+ of GMV, which refocused stock and planning priorities for the category teams.
5. **Category analytics suite (Noon):** Built OOS, price-competitiveness, assortment-health and P&L (NMV/PC1) tracking for 12 AMs across KSA and UAE, on BigQuery, Airflow, Looker Studio and automated alert mailers.
6. **Customer segmentation (Citibank):** Built a segmentation model that drove 84% of sales.
7. **Claims-cost prediction (UnitedHealth Group):** Built an XGBoost model predicting healthcare claims cost at AUC 0.84, deployed on Azure ML.
8. **Fraud detection and NLP (UnitedHealth Group):** Built Isolation Forest fraud detection and an NLP document-classification model to automate claims review.
9. **Experimentation and churn (Eucloid):** Built A/B testing infrastructure and a churn prediction model for SaaS product teams.
10. **Job Radar (personal AI product):** Built an automated 9-stage LLM pipeline that scrapes 51+ job sources, scores roles with Claude Haiku, and tailors resumes and cover letters with Claude Sonnet, running daily on GitHub Actions.