Science leader in CAAI (Cloud Applied Artificial Intelligence) that enables AI/ML in next-generation data center end-to-end lifecycle management. My team is actively developing productionized solutions in:- Critical Environment. Anomaly detection system that monitors MS data center devices and identifies anomalies- Technician Management. Task Classification System that leverages NLP to map the ticket description to pre-defined task categories. - Construction. Schedule as a Service to enable prediction, prescription and optimization of construction schedule. - Safety. Proactively identify potential safety incidence.

- Designed scalable algorithms to estimate long-term value for high-value actions. Advocated company-wide applications and adoptions of long-term value metrics in 10+ organizations and systems. - Contributed to the design and implementation of an automated system that optimally price 300K Private Brand products that generate 300M long-term value annually. Achieved 7% lift in this long-term metric during A/B test. - Led the science effort of an automatic system that recommends the optimal cost target for all Amazon retail vendor negotiations. Reduced total cost of negotiation by 5% in pilot study. - Acted as a thought leader in various business space and interact cross-functionally with a wide variety of leaders and teams to identify high-impact opportunities and drive long-term science roadmap. Scoped S-level goals and advised senior leadership on topics including investment and selection strategies for Private Brand and Consumable business, scientific technology choices and dynamic seller inbounding fees.

- Led a team to design and automate a system that estimates the driver pool capacity and recommends the precise number of drivers to recruit. Realized $2.5M YoY cost reduction and a 5% increase in engagement metric during peak 2018 without compromising fill rate. - Led a team to design and implement a machine learning system that uncovers personalized working preference for Amazon delivery partner. Incorporated the output into Last Mile work scheduling system and achieved a 24% increase in exclusive offer fill rate. Patent pending.

- Designed an automatic forecasting platform to predict AWS network traffic for 250K+ network spans in 12-18 months horizon. Controlled forecast error within 2% for major network layers. Patent pending. - Identified network bottleneck by examining network effective capacity and traffic prediction. Drove the spending of two-billion Network Scaling budget to minimize idle capacity and maximize performance.

- Constructed aggregate level demand forecasting model for ebook units. Controlled the forecasting error within 2% in 2013 Q4 for US and within 1.5% for UK marketplaces. - Developed the framework to evaluate short to medium-term impact of merchandising activities on customers’ life time value. Achieved 50% increase in medium-term customer engagement metrics through targeted campaigns.