Worked closely with managers and engineers, performed data analysis, revised infrastructure, conducted experiments, devised model evaluation, mostly in Python and Scala, to improve Etsy’s ranking recommendation system, within a team of 7
Initiated an project to improve model relevance, found pipeline weakness, introduced evaluation metrics, designed methods boosting recommendation relevance, optimized learning efficency, reduced model bias, increased weighted NDCG metrics by 15% averagely, strongly mitigated over 70 persistent problematic queries
Participated in projects to alternate model price sensitivity, tested different approaches including prioritizing personal price preference and prioritizing high price listings, improved conversion rate by 2%
Updated feature pipeline by adding new features and located obsolete features, tested new feature engineering model impact
