Senior Recommendation Systems Engineer – Real-Time Serving (Full Remote)

Overview   We are a fast-growing social and livestreaming platform — think Facebook/X meets live video — serving millions of users globally. We're looking for a senior engineer to own our real-time recommendation serving system: from candidate retrieval and ranking to personalized delivery at scale.   You'll be inheriting an existing system and driving its evolution — moving from batch-based to real-time architecture — not starting from scratch.   ---   What You'll Do   - Lead the evolution of the recommendation online serving layer: real-time candidate retrieval, multi-stage ranking, quality filtering, and personalized delivery   - Drive the migration from batch-based to real-time recommendation serving   - Build and maintain feature engineering pipelines for users, content, and creators   - Collaborate on the model lifecycle from offline training to online serving and iteration   - Collaborate async with engineering, product, and data teams across timezones   ---   Requirements   - Hands-on production experience with real-time recommendation serving systems — social media or content platform background strongly preferred   - Strong proficiency in Python, SQL, Kafka, and Redis   - Ability to read and work with Apache Spark-based pipelines   - Experience with real-time model serving, low-latency inference, and production reliability   - Strong system design skills; able to deliver end-to-end solutions independently   - Fluent Chinese (native or near-native) — primary working language for internal communication and team collaboration; English proficiency sufficient for reading technical documentation   ---   Nice to Have   - Experience with Flink, Databricks, or Elasticsearch   - A/B testing, experiment design, and data analysis experience   - Go, Rust, or TypeScript experience for high-performance backend serving   - Vector search (Faiss, Milvus, Pinecone)   - Livestreaming backend infrastructure (HLS, RTMP, WebRTC)   - Ad systems or monetization recommendation experience   - Experience using modern AI-assisted engineering workflows is a plus   ---   Work Setup   - 100% remote   - Asia timezone preferred (UTC+7 to UTC+9)   - Async-first collaboration with defined working hours; on-call expectations will be discussed during interview