Qingwei Li

Customer Segmentation for Targeted Advertising Recommendation

• Utilized Python to implement unsupervised learning algorithms to segment around 1 million active users based on user's features, enabling the marketing team to generate targeted advertising.
• Contributed to $86,000 in savings during 2021 Q3 by deploying a K-Means clustering model for user segmentation, which sharpened the marketing team's targeted advertising strategy and optimized campaign budgets.
• Collaborated closely with cross-functional teams to integrate clustering algorithms into the marketing pipeline

Image Classification and Model Compression Acceleration

• Built a computer vision system for product inspection by implementing object detection models. Achieved a 90% inspection accuracy rate.
• Performed model acceleration using the mixed precision training. Implemented the trained model for image de-noising, object detection, and product quality classification on conveyor belts and filtered out defected items.
• Contributed to 32% model training time reduction by implementing knowledge distillation and the mixed precision training. Deployed the model and saved 10+ hours of human labors per week for product inspection.
  

• Collaborated with the German engineering team on data acquisition and data processing, focusing on optimizing the
  performance of machine vision systems used in indoor 3D scanning devices.
• Analyzed customer feedback to identify potential areas of improvement of 3D scanning devices. Proposed and
  implemented strategies to enhance the accuracy and efficiency of the devices, leading to increased customer satisfaction.
• Assisted in the selection and calibration of cameras, lenses, and lighting for the 3D scanning devices, ensuring optimal
  image acquisition and processing for scanning and purposes.