Nan Su

Data Scientist with 10+ years of experience developing and deploying data-driven solutions using machine learning, causal inference, experimentation, and statistical modeling. Expertise in regression, decision trees, neural networks, time-series analysis, clustering, causal discovery/inference, and A/B testing. Proficient in Python, SQL, PySpark, TensorFlow, PyTorch, AWS, cuDF, Azure, Databricks, and R. Multiple professional certifications in LLM development using AWS Bedrock and Agentic AI.

Kaggle Competition

Expert (Top 1%).

Data Scientist - Amazon - Seattle, WA

(2022-07)

• Developed an end-to-end causal inference pipeline combining automated causal-graph discovery and double machine learning (DML) effect estimation to quantify operational drivers of productivity, delivering $19.5M in annual cost savings.
• Partnered with engineers to build LaunchGate, a multi-agent GenAI solution that automates experiment design and analysis; applied it to quantify the impact of stow baskets on productivity across the AR-sortable network.
• Developed an LLM-powered science solution using Claude Sonnet 4 for item disposition determination, improving decision accuracy while delivering 2.7% cost savings, 21-second processing-time reduction, and 0.8-second AI latency.
• Developed a two-phase ML modeling approach to detect broken sets, reducing warehouse damages and customer concessions and delivering $11.5M in annual cost savings.
• Developed and productionized ML models to automate inventory ramp-up planning for new fulfillment centers across the US and Canada, delivering $4.5M in annual labor and capacity-planning savings.

Data Scientist - HCL Technologies - Redmond, WA

(2020-10 - 2022-06)

• Developed time-series forecasting models to predict customer service call volumes and support Microsoft's customer service strategy.
• Built a cloud-based change-point detection solution to identify shifts in customer service demand and improve forecasting.
• Developed ML models to predict service bug closure timelines under a new SLA standard, enabling the C+AI team to prioritize issues and accelerate resolution.

Assistant Director - Liberty Mutual Group - Boston, MA

(2015-06 - 2020-09)

• Implemented anomaly detection techniques to automate fraud detection and prevention for the Special Investigation Unit.
• Developed deep learning models to analyze customer feedback and extract sentiment-bearing phrases for customer experience insights.
• Applied propensity score matching to quantify the financial impact of medical treatments and identify opportunities to improve outcomes.
• Defined driver-behavior and claim-frequency metrics using telematics data processed on AWS EMR.

Sr. Consultant - The Travelers Companies - Hartford, CT

(2012-11 - 2015-05)

• Developed predictive models to forecast insurance losses using diverse data sources, including third-party appraisal, text, and medical bill data.

Ph.D. - Statistics - Clemson University (2012-05)

B.S./M.S. - Applied Mathematics - Zhejiang University (2006 - 2007)