● Debug C#/.NET Core services (BBGAIAgentLib, AIAgentWeb) to validate Azure AI agent threads and ensure proper serialization/deserialization of multi-turn messages across users, supporting persistent memory and reliable orchestration logic. ● Build CI pipelines using Azure DevOps to automate .NET build/test workflows and lay the groundwork for downstream deployment of user-facing features like thread search via asynchronous APIs and streaming output.

● Segmented 1,000+ B2B retail accounts into six behavioral groups (e.g., high-volume loyalists, price-sensitive churn risks) to inform upsell and retention strategies, enabling targeted sales planning across the commercial team. ● Built 30+ revenue-weighted and categorical features to represent retailer behavior, applied PCA with component attribution for dimensionality reduction, and used KMeans and Agglomerative clustering (Silhouette 0.61) to define interpretable segment. ● Presented data-driven revenue uplift insights to C-suite stakeholders based on retailer segmentation results, and proposed a tiered upsell campaign roadmap to boost repeat order rates across high-value segments.

● Increased web event tracking accuracy by 11% ahead of a global e-commerce promotion by auditing and correcting tag discrepancies between implementation guides and live tracking data, enabling cleaner downstream analytics. ● Automated country-specific data retention in BigQuery using scheduled SQL queries, managing 150+ daily-ingested datasets across 30+ global entities and reducing manual data purging time by 40% in line with client compliance policy. ● Streamlined and standardized 20,000+ tracking tags across 100+ global e-commerce sites, reducing deployment errors by 78% in Adobe Experience Platform and Google Tag Manager and improving tracking consistency across platforms. ● Designed a GDPR-compliant SQL data pipeline used by 50+ employees to enforce regulatory compliance in Europe.

● Engineered a Python pipeline to canonicalize company names in the database by resolving multilingual and alias variations, reducing manual review time by 75% and improving entity-level accuracy to 99.8%. ● Developed a Selenium-based scraping framework to collect and enrich missing company data for integration into internal platforms, streamlining downstream workflows and enabling 20+ team members to populate Snowflake with complete records.