Computer Vision Algorithm Engineer - TikTok E-commerce Knowledge Graph

岗位职责 / Responsibilities

Our team is responsible for developing state-of-the-art CV/NLP/ML algorithms and strategies to improve user consumption experience, inspire merchants' service quality and revenue, and build a fair and flourishing ecosystem on our E-commerce Platform. More specifically, our team is responsible for the algorithms of Product Knowledge Graphs under TikTok's global e-commerce business.

What you will do:

• Participate in the development of massive knowledge graphs of real-world products to support feed ranking, recommendations, and ads.
• Collaborate with product managers, data scientists, and the product strategy & operation team to define product strategies and features.

Responsibilities:

• Knowledge graph construction, including product/content/feedback understanding and category/brand/SPU construction.
• Explore the implementation of Knowledge Graph and Cognitive Graph on the e-commerce side of TikTok e-commerce shopping guide.
• Responsible for the mining of e-commerce shopping guide knowledge such as shopping scenes/people/goods matching/product layering.
• Responsible for the optimization and iteration of computer vision related models in the e-commerce scene, including fine grain classification, product object recognition, product subject recognition, feature extraction, logo detection, brand recognition, etc., to optimize the merchant's product loading and unloading process.
• Responsible for e-commerce short video and livestream classification, multi-modal content mining, multi-modal content understanding, optimize e-commerce short video and livestream shopping experience.
• Responsible for e-commerce image search, photo search goods, goods duplication algorithm.
• Explore the cutting-edge technology of computer vision, responsible for the iteration and evolution of the overall algorithm and system.

任职要求 / Requirements

Minimum Qualifications:

• In-depth knowledge in a certain field of multimedia and computer vision, including but not limited to: image search, image/video classification and recognition, image segmentation, object detection, OCR, graph neural networks, multimodal, unsupervised and self-supervised learning, etc.;

Preferred Qualifications

• Familiar with the training and deployment of one or more framework models of TensorFlow/PyTorch/MXNet, and understand training acceleration methods such as hybrid precision training and distributed training;