Experience

Aibee Inc.

• China
• Software Development
• 1 - 100 Employee

• Research Engineer

  • Jul 2019 - Present

  ● Researched the state of art papers on detection and implemented the algorithm with PyTorch and TensorFlow, made quantitative models and released NNIE usable models for shopping malls and store camera.● Filtered large number of images with masks from database and trained the mask quality model to make facial recognition equipment work normally during the Covid-19 epidemic.● Developed the module in critical data processing project with C++ which predicts the reasonable false negative detection boxes and adds them back, bringing 20% increase for recall.● Improved the scheduling script for data processing project with Shell and Python to reuse loaded models repeatedly for reducing processing waiting time. ● Designed task schedule system with Airflow for submitting data processing task to Kubernetes cluster which in the customer owned servers. ● Made a toolkit based on web with Flask, helping deep learning training on GPU cluster more friendly-used. It supports Maskrcnn Benchmark, Detectron2 and other popular frameworks using in the group.



Aibee Inc.

• China
• Software Development
• 1 - 100 Employee

• Research Engineer

  • Dec 2017 - Apr 2019

  ● Reeded related papers with tracking algorithm, evaluated with our own datasets, and built the tracking algorithm verification demo ● Participated in the development of web-based labeling tools and deploy labeling tasks with docker for use by labeling teams (over 200 people) in different cities. ● Based on the mask-rcnn detection network, improved the network structure and fine-tuned the parameters to provide detection models for hundreds of cameras within Chow Tai Fook stores and Wanda shopping malls.