Projects
Operational AI product portfolio
Senior Product Manager building Operational AI Systems at Amazon scale — combining systems thinking, technical prototyping, human-in-the-loop workflows, and measurable business outcomes.
Shipped Products
Marketplace catalog systems
Shipped ProductProduct Variation Intelligence
Unlocked large commercial upside while balancing automation with review controls.
ML clustering + LLM validation + confidence scoring to improve product family and variation quality across categories.
Checkout + delivery experience
Shipped ProductPin-on-Map Address Creation (India)
$835K annualized cost savings, geocode precision uplift, and conversion improvements.
Customer address creation flow using map pin + form fallback; relaunched after infra and UX fixes.
Global delivery reliability
Shipped ProductTargeted Address Intervention at Checkout
Reduced delivery failures and refund contacts while protecting core guardrail metrics.
Intervention widget prompting customers to correct defective addresses across 8 country stores.
Revenue recovery
Shipped ProductNo-Saved-Address Checkout Recovery
~$400M annualized incremental gross sales and large reductions in support contacts/refunds.
Guided flow to recover customers with no saved address at checkout, scaled across MENA stores.
Self-built AI pipeline
Shipped ProductLLM Product Summary Regeneration
+1.7% gross sales per session in A/B test and roadmap path to 257+ product types.
PM-built end-to-end LLM pipeline publishing 40K+ regenerated summaries and handed to science for scale-up.
Catalog content quality
Shipped ProductAI-assisted Title Quality Engine
10M+ titles enriched, title quality lifted to 93–96%, and $8M annualized gross sales impact.
Arabic-first title quality engine that enriched 10M+ titles and improved quality scores across categories.
Growth and discovery
Shipped ProductHomepage Discovery Launch
$37.5M annualized gross-sales uplift and 8 additional features launched at zero engineering cost.
Localized infinite-scroll personalized discovery and unblocked launch economics before Ramadan peak.
Operational AI Systems
E-commerce content quality
Operational AI SystemProduct Detail Page Quality Auditor
Created a repeatable quality workflow for merchandising teams with auditable outputs and prioritized remediation.
Playwright + S3 + Claude multimodal system to audit Arabic mobile product detail page quality with evidence-backed issue labels.
AI reliability and evaluation
Operational AI SystemHallucination Audit System
Enabled reliability dashboards that informed prompt tuning, policy thresholds, and human review prioritization.
Breaks generated catalog content into claims and labels each against source context: Exact Match → No Match.
AI-assisted decision support for the final moments before delivery
Operational AI SystemDelivery Intelligence Platform
Demonstrated how operational AI can bridge the gap between map navigation and real-world delivery execution, and established a reusable framework for multimodal AI in high-friction workflows.
Explored how multimodal AI can help drivers navigate the most failure-prone stage of delivery: the final 100 meters before reaching the customer.
Experimental AI Systems
Personal emergency intelligence system
Experimental AI SystemDubai Alert Tracker
Delivered timely alerts for major events (interception confirmations and airline suspension signals) while reducing false alarms through multi-source consensus gating.
Built during March 2026 regional attacks from a phone-only setup to convert high-noise Reddit megathread activity into verified community-signal alerts.
Personal AI project
Experimental AI SystemThreadCast
Demonstrated how AI can transform discourse into structured, reusable media artifacts.
Converts Reddit threads into multi-persona podcast scripts and audio, including debate format and cliffhanger logic.
AI-powered property intelligence and investment decision support
Experimental AI SystemPropertyGPT
Demonstrated a practical AI workflow for real-estate decision support, turning scattered market signals into explainable, user-specific insights.
PropertyGPT explores how AI can assist buyers and investors by combining listing data, location intelligence, infrastructure signals, rental yields, market trends, and user preferences.