2019年11月 - 现在
. Worked with a team of 5-6 engineers, design and development E2E workflow for 15+ computer vision-related projects and reduce 50% of the AI project lifecycle.
. Build AI model by transfer learning and fine-tuning and training deep neural networks by Neural Structured Learning with deep learning frameworks.
. Develop TensorFlow input pipelines by using SQLite and Structuring TFRecords for convert images into the TFRecord format, to use in deep learning algorithms.
. Create and design image datasets by mariadb and called by MySQL from backend, also automated metadata with a CronJob on Kubernetes.
. Set up Node-exporter and Prometheus to collect metrics of the Linux hosts and create dashboards in Grafana to visualize and monitor the metrics.