Career

Leading development of Real-time Next Best Page Recommendation System, a recommendation engine for QuickBooks.com. Designed a two-stage architecture pairing an LLM-based relevance-analysis agent with a causal uplift model for conversion scoring, trained and validated on historical clickstream and conversion data. A/B testing showed positive lift across signup and GNS, driving an estimated ~$20M annual incremental LTV, with the model now deployed across 100% of US desktop traffic. Also co-architected SINGULAR, an internal MCP server that unifies data knowledge across every domain into one governed layer, paired with Intuit's Databricks MCP so agents go straight from knowing what to query to executing it. Filed a patent application for SINGULAR, now broadly adopted by DS, marketing, and product teams org-wide to run complex analysis within Claude, Cursor.

Responsible for full-stack development of machine learning models and generative AI applications, as well as operational data analysis, providing high-quality solutions to customers in various industries.

Owned the data science work behind creator growth, retention, and monetization on Kuaishou's client operations team, spanning four workstreams: (1) built predictive models to forecast creators' potential livestreaming earnings and designed tiered incentive strategies, with A/B testing showing a $5M monthly revenue increase; (2) partnered with operations to reduce creator churn through feature engineering and SHAP analysis, identifying key drivers including early-stage follower growth, then designed and validated retention strategies that reduced long-term churn by 20% and became standard practice; (3) collaborated with top influencers (KOLs), applying propensity matching and difference-in-differences (DID) analysis to isolate the causal effect of their content on engagement, then used uplift modeling to quantify heterogeneous effects across user segments, revealing the strongest impact among high-value male users and supporting a $10M increase in KOL partnership investment; (4) owned A/B test design and real-time monitoring for new user and creator strategies, applying quasi-experimental methods, including synthetic control and matching, in settings where randomization was infeasible.