WXG - Data Scientist
Description
WeiXin Group(WXG) is responsible for the construction and operation of the Weixin ecosystem and leveraging Weixin's open platforms such as Official Accounts, Mini Programs, Weixin Pay, WeCom and search function. WXG provides solutions and connectivity for intelligent upgrades across all industries. WXG is also responsible for the development and operation of QQ Mail, WeRead and other products.
Responsibilities
- Explore both qualitative and quantitative data sets, perform in-depth analysis to derive actionable insights, identify problems, and propose data-driven solutions for the WeChat ecosystem
- Collect large sets of structured and unstructured data from independent sources; clean, validate, and transform data to ensure accuracy, completeness, and uniformity
- Design and maintain dashboards, reports, and automated data pipelines to support ongoing business monitoring and decision-making
- Analyze data to identify patterns, trends, and anomalies, uncovering potential opportunities and/or threats to the WeChat ecosystem
- Interpret findings to discover solutions or opportunities and build story-boards to communicate insights to the business for action
- Own and evolve the computer vision infrastructure, including building, optimizing, and maintaining CV model training/serving pipelines, evaluation frameworks, and related toolchains
- Develop and deploy computer vision solutions (e.g., image/video understanding, OCR, content moderation, multimodal analysis) to address business needs across the WeChat ecosystem
- Leverage LLM-based approaches and other machine learning / deep learning techniques where appropriate to enhance analytical capabilities and support intelligent product features
Requirements
- Bachelor's degree or above in Computer Science, Artificial Intelligence, Mathematics, Statistics, Data Science, or a similar quantitative field
- Hands-on experience with LLM application development (e.g., fine-tuning, RAG, Agent frameworks such as LangChain/LlamaIndex/AutoGen) is strongly preferred
- Experience with computer vision models (e.g., object detection, image classification, multimodal models) is a plus
- Good knowledge of Python; familiarity with C/C++ is a plus
- Familiarity with mainstream deep learning frameworks (PyTorch, TensorFlow) and model deployment tools (e.g., Docker, FastAPI, TensorRT) is preferred
- Experience with vector databases (e.g., Milvus, FAISS, Pinecone) or data pipeline tools is a plus
- Ability to communicate complex technical ideas to non-technical stakeholders in a clear and accessible manner
- Self-driven, innovative, and collaborative, with strong communication and presentation skills; able to bridge business and technical audiences
