Bingyu Shen & Rohan Mahadev
Pinterest Labs
Tuesday, October 4, 2022
3:30 pm - 4:30 pm
Zoom
Title: Quality Content at Pinterest
Abstract:
Measuring quality of contents and limiting the distribution of low quality contents is important to maintain a good user experience for Pinterest. In our team we develop various signals to measure the quality of content users created using deep learning / machine learning techniques such as graph neural networks, Bert, etc. The main goal of our team’s job is to efficiently serve high quality pins responding to users’ requests. In this talk we will share our work experiences as part of the content quality team at Pinterest.
Bingyu Shen is a software engineer at Pinterest. He graduated from the University of Notre Dame with a Ph.D. in Computer Science. His research focuses around text understanding and text retrieval in the wild. Bingyu has been with Pinterest for 1 year and continues to use his knowledge in NLP and Computer Vision to solve related problems in the real world.
Rohan Mahadev is an ML engineer on the Advanced Technologies Group at Pinterest where his work on computer vision focuses on inclusive + creative AI applications of computer vision along with large-scale distributed multi-task learning. Some of the technologies his group built recently were the Skin Tone and Hair Pattern filters for search and diversification of content and the Shuffles app which lets users make creative collages from the content they find on Pinterest. Prior to joining Pinterest, Rohan was a grad student at NYU where he worked on self-supervised multi modal training of models and fairness/ethics in machine.
