Applied Data Scientist

Applied Data Scientist — Aether Analytics

Seattle, WA | May 2023 – Present

• Developed and deployed a real-time gradient boosted decision tree (GBDT) model using XGBoost to predict customer churn, reducing churn rate by 17.5% ($1.2M ARR saved) in 4 months.
• Architected a multi-armed bandit recommendation pipeline in PyTorch, serving 12M daily active users and increasing click-through rate (CTR) by 22% over a 3-month trial period.
• Designed and executed 45+ automated A/B tests on search query routing algorithms, leading to a 6.4% improvement in search-to-purchase conversion rates.
• Containerized 14 legacy machine learning models using Docker and Kubernetes, reducing compute latency by 35% and infrastructure costs by $180,000 annually.

Data Scientist — Helix Tech Group

Seattle, WA | June 2021 – April 2023

• Built a random forest classification model to detect anomalous payment transactions, improving precision from 84% to 96% and reducing false positives by 40%.
• Engineered ETL and feature engineering pipelines in Apache Spark and SQL, processing 4TB of weekly telemetry data and reducing data ingestion times by 28%.
• Partnered with 3 product owners to translate business requirements into statistical metrics, decreasing customer ticket response times by 15% via automated routing models.
• Optimized hyperparameter tuning pipelines using Optuna, saving 22 hours of GPU training time per model release cycle.

Machine learning, Deep learning, Statistical modeling, A/B testing, Feature engineering, Natural Language Processing (NLP), Python, SQL, PyTorch, TensorFlow, XGBoost, Scikit-learn, Apache Spark, Docker, Kubernetes, AWS (S3, SageMaker, EC2), Git, CI/CD

Predictive Customer Churn Engine

Role: Lead Data Scientist

Tools: Python, XGBoost, AWS SageMaker, SQL

Built an end-to-end churn prediction service that identified at-risk users with 91% accuracy, allowing targeted retention campaigns that saved $320,000 in customer acquisition costs.

Telemetry Feature Pipeline Optimization

Role: Applied Data Scientist

Tools: Apache Spark, Docker, Kubernetes, AWS EMR

Designed a distributed feature extraction pipeline handling 10B monthly events, reducing training data compilation time from 5 days to 14 hours.

• AWS Certified Machine Learning – Specialty (2024)
• TensorFlow Developer Certificate (2022)

• Languages: English (Native), Mandarin (Conversational)
• Volunteer Work: Technical mentor for high school data science outreach program (2023-present)
• Availability: 3 weeks notice