Expert Backend Engineer - Machine Learning Parameter Server (EGO Team)

About the Team

The EGO team is committed to creating a top-tier machine learning platform designed to efficiently support algorithm deployment across diverse business areas, including recommendation, search, and advertising. The team's work centers on achieving exceptional optimization for CTR and CVR predictions within large-scale sparse parameter environments, maximizing e-commerce platform performance and delivering enhanced value to the company.

This platform encompasses the entire deep machine learning lifecycle—from data sample management through training, model creation, deployment, online loading, and inference services. It features an intuitive Web UI and a RESTful API that allows a seamless end-to-end machine learning experience.

Job Responsibilities

• Design and implement distributed Parameter Server (PS) systems aimed at supporting large-scale sparse model training and inference across domains like search, advertising, and recommendations.
• Ensure the system can manage extremely high-throughput parameter operations—reading, writing, and updating—with scalability to hundreds of billions of features and TB-scale sparse models.
• Enable real-time online learning, addressing algorithmic requirements such as feature admission and expiration effectively.
• Contribute to the advancement of the integrated machine learning platform by embedding the PS system to offer a robust, user-friendly, high-performance distributed service for parameters.
• Improve platform efficiency and usability to speed up model iteration cycles for algorithm teams.

Qualifications and Experience

• Possess a Bachelor's degree or higher in Computer Science, Electronics, Automation, Software Engineering, or related disciplines, with a minimum of six years' relevant experience.
• Demonstrate mastery in C++ programming, especially in low-level technical domains including multi-threading, lock and memory pool optimization, thread pool implementation, template programming, debugging using GDB, performance enhancement, and RPC framework development.
• Experience with distributed Parameter Server architectures, backend optimizations for distributed systems, high-performance in-memory key-value storage systems, NVMe-SSD based KV storage, and efficient client-server systems will be advantageous.
• Exhibit strong passion for computing technologies, a proactive attitude toward learning, thoroughness in research and practical implementation, and uphold rigorous standards and precision in code quality.
• Be a collaborative team member with an eagerness for continuous personal and professional growth.