Official platform of JD Retail Technology, delivering insightful R&D news and a deep look into the lives and work of technologists.
This article details JD Retail's award‑winning, data‑driven inventory selection and allocation algorithms, explains their mathematical models, heuristic and end‑to‑end learning solutions, presents experimental results on real‑world data, and quantifies the operational gains achieved after deployment.
MODRL‑TA is a multi‑objective deep reinforcement learning framework that unites independent Q‑learning agents, a cross‑entropy‑based decision‑fusion module, and progressive data‑augmentation to overcome cold‑start and multi‑objective trade‑offs in e‑commerce traffic allocation, delivering up to 18% more impressions, 4% higher CTR and 5% higher CVR in live tests.
This article analyzes JD Retail's approach to automating AB experiment workflows by introducing a data‑weaving framework that unifies metric definitions, streamlines logical data modeling, and enables scalable, real‑time DAG orchestration across multiple experiment scenarios.
JD.com’s data‑governance framework combines a health‑score‑driven, automated platform that cross‑verifies audit logs, builds full‑link and operator‑level lineage, introduces standard fields, and optimizes resource mixing, task staggering, and cross‑datacenter scheduling, while targeting real‑time AI‑enhanced detection and full automation.
This article introduces JD’s self‑developed HarmonyOS image library, explaining the background, research on existing solutions, the decision to build a C++‑based cross‑platform library, its modular architecture, performance optimizations, quality assurance mechanisms, and future development plans.
This article presents a comprehensive case study of how JD's procurement system (BIP) tackled massive data volume and complex query challenges by redesigning data models, introducing heterogeneous storage for inbound orders, leveraging JED and JimKV, and offloading complex searches to Elasticsearch, resulting in dramatically reduced database load and improved user experience.
This article presents a comprehensive study of generative retrieval in large‑scale e‑commerce search, detailing lexical‑based and SemanticID‑based methods, their challenges such as long‑tail distribution and token length, experimental evaluations, the discovered "sandglass" effect, and proposed solutions to improve recall and efficiency.
This article explains how Taro leverages C++ and the HarmonyOS C‑API to build a high‑performance, multi‑language event system that bridges ArkTS, JavaScript and native components, detailing architecture considerations, implementation details, and future directions for improving cross‑language event handling.
This article explains why Java code style standards are essential for team collaboration, reviews major guidelines such as Google Java Style and Alibaba's p3c, and provides step‑by‑step instructions for configuring IntelliJ IDEA, Checkstyle, and Maven to enforce consistent, readable, and maintainable Java code.
This article documents the diagnosis of slow SpringBoot startup in a large‑scale advertising platform and presents a series of optimizations—including actuator monitoring, Tomcat TLD scan disabling, HBase async warm‑up, custom BeanPostProcessor timing, asynchronous JSF consumer refer, Tomcat version tuning, and hardware migration—that together reduce launch time by about 60%.
