Machine Learning Engineer, E-commerce Governance Algorithms

About the Team- Shaping the Future of Global E-Commerce Experience

We are the Governance & Experience Algorithm Team — a trailblazing AI force at the forefront of building a trusted, efficient e-commerce ecosystem. Established in 2020, we leverage Graph Neural Networks, Multi-Objective Optimization, Time Series Prediction, and Large Language Models (LLM) to tackle some of the most complex challenges in e-commerce:

Quality Revolution Combat low-quality products and fraudulent practices across millions of listings:

• Detect low-quality items, counterfeit goods, and false advertising.
• Identify "not-as-described" discrepancies between product claims and actual quality.
• Flag risky sellers through comprehensive quality assessments.

Fulfillment & Logistics Experience Reinvention Enhance delivery reliability and operational efficiency:

• Resolve delivery delays and non-receipt issues.
• Prevent inventory shortages via predictive analytics.
• Optimize warehouse operations for faster order fulfillment.

Why Our Work Matters:

• We protect users globally, ensuring safe shopping experiences in international markets and beyond.
• Our innovations power TikTok’s mission to “Inspire Creativity, Bring Joy” by fostering trust and delight in every transaction.
• We’re not just solving problems—we’re redefining industry standards for e-commerce product, service, and logistics experiences.

What You’ll Do

1. Lead Cutting - Edge AI Projects - Build graph-powered risk networks to uncover similar product clusters and high-risk seller/creator groups in emerging markets, reducing false advertising incidents in global regions. - Develop LLM-based multi-modal systems using cross-modal fusion (text, images, behavioral data) to detect false advertising and low-quality products. Deploy AI-powered product business suggestions to improve seller compliance and user trust. - Lead time-series forecasting innovation for logistics performance metrics (e.g., delivery delays, cancellation rates), driving improvement in the e-commerce experience through predictive analytics.

2. Solve Complex Business Challenges

• Develop multi-task models using MMoE and dynamic loss functions, driving measurable improvements in platform product, service, and logistics health.
• Transform supply chains by optimizing warehouse product quality qualification strategies to cut missing recalls and reduce warehouse costs.
• Unleash RPD/RPR growth through causal inference frameworks, identifying hidden correlations between CCR and user behavior to design personalized recommendations.

1. Redefine Industry Standards

• Optimize LLM reasoning with DPO/GRPO to enhance fraud detection, outperforming traditional SFT methods. Enhance tabular data modeling for better explainability in e-commerce risk assessments.
• Construct heterogeneous graphs modeling product - seller - creator - video - user relationships, enabling improvements in product lifecycle insights and detection of similar product patterns.
• Build unified time-series forecasting models across products, sellers, creators, videos, and users, achieving SOTA performance in predicting inventory shortages and demand surges.