Official account of DaTaobao Technology
The article traces the evolution of AI in software development, explains why traditional code‑quality processes struggle with Blocker issues, and details how a browser‑automated AI SRE agent can discover, route, and fully repair these problems while keeping humans in the final review loop.
The article details how the product team tackled AI‑driven challenges by designing a two‑layer, event‑driven Function‑Centric Agent architecture that unifies workflow orchestration and capability supply, enabling real‑time inference for billions of items, cutting development cycles to one person‑week, and boosting search conversion rates.
MNN‑Sana‑Edit‑V2, a collaborative effort between Taobao’s Meta team and Hangzhou University, combines a frozen Qwen3‑0.6B LLM, Learnable Query, Connector, Linear DiT and Deep Compression Autoencoder with 4/8‑bit quantization to run fully on mobile devices, delivering 512×512 comic‑style conversions in about 15 seconds—2.5× faster than cloud alternatives—while providing open‑source code, detailed training stages, and extensive performance benchmarks.
This article details the AI-driven testing platform built for Taobao’s marketing venues, describing how large‑language models and multimodal agents enable visual rendering verification, price and content consistency checks, and automated multi‑device adaptation, resulting in a 40% overall efficiency boost and a 100% increase in tester productivity.
The article explains how a natural‑language‑driven AI platform, with pre‑configured agents and parameterized calls, can replace a complex 150‑line SKU‑combination risk‑checking algorithm in e‑commerce, dramatically improving readability, maintainability, development speed, and business value while reducing cost and response time.
The transaction terminal team created an AI‑driven workflow that automatically collects, cleans, classifies, alerts, distributes, attributes, and reviews user feedback, using a four‑step LLM model to ensure controllable, consistent, and explainable sentiment analysis while boosting efficiency and trust.
This article presents the Trustworthy Generation design pattern for Retrieval‑Augmented Generation (RAG) systems, analyzes four root causes of low trustworthiness—retrieval errors, content reliability, pre‑retrieval reasoning mistakes, and model hallucinations—and proposes layered solutions, citation techniques, CRAG and Self‑RAG architectures, guardrails, and practical trade‑offs.
This article records the end‑to‑end practice of using the AI coding tool Cursor to deliver the front‑end of a red‑packet sign‑in page’s "single‑product rebate" feature, covering requirement breakdown, AI‑generated design and code, popup animations, feed insertion, deduplication, debugging, and final verification.
This article presents DeepStock, a deep reinforcement learning framework with policy regularization that integrates classic inventory heuristics, achieving 7% turnover reduction and multi‑million cost savings across millions of SKU‑warehouse pairs in Alibaba's self‑operated ecosystem.
This article proposes an ASR‑driven pipeline that creates high‑quality AI‑reply vs. human‑like reply pairs, trains a rewrite model and a reward model, and uses GRPO reinforcement learning to generate natural, helpful, and less AI‑sounding responses in digital‑human live streaming, achieving 92% accuracy and 97% helpfulness while improving user experience.
