Job Description

About the Team:

Search is no longer just a feature—it is the defining battleground that will determine whether TikTok Shop becomes the first truly global e-commerce super-app. Our Search platform is the core engine behind the "shelf-field" (product card) strategy, and the key to competing head-to-head with Amazon, Taobao, and Pinduoduo in a global market worth over $10 trillion.

Search will propel TikTok Shop from tens of billions to hundreds of billions in annual GMV by activating real shopping intent, unlocking zero-sales products, and ensuring every item becomes instantly discoverable the moment a user types or taps.

We innovate where extreme performance, massive global scale, and uncompromising consistency meet.

Responsibilities:

You will design, build, and optimize the core infrastructure that supports TikTok Shop’s recall, ranking, and re-ranking pipelines globally.

- Core Search Engine Development: Design and implement high-performance online retrieval systems. Optimize core components including the inverted index, vector retrieval (ANN/HNSW), query understanding, and merger logic.

- Real-Time Data Pipelines: Build highly scalable and fault-tolerant data pipelines using Flink, Kafka, and Spark to ensure product changes (price, stock, and new listings) are reflected in search results in near real-time.

- System Stability & Performance: Drive latency, throughput, and cost optimizations for services handling hundreds of thousands of QPS. Troubleshoot complex distributed system issues, manage cross-region failover, and design high-availability disaster recovery solutions.

- Large-Scale Storage & Retrieval: Design and optimize distributed storage libraries (based on RocksDB/Redis) and columnar databases tailored for high-speed e-commerce feature retrieval.

- ML Infrastructure Collaboration: Work closely with Algorithm/ML Engineers to productionize state-of-the-art Large Language Models (LLMs) , AI Search, and multi-modal search models, ensuring the infrastructure supports massive model serving and real-time feature engineering.