Internships


NVIDIA Deep Learning Algorithm Team (May 2022 - August 2022)
  • I worked in a team of deep learning algorithm engineers and NVIDIA researchers during my internship. I developed a new hierarchical graph neural network model leveraging a cross-attention mechanism for entity resolution tasks. Specifically, I developed the model for the cross-device user matching problem, and the model achieves significantly better performance (5%) compared to the state-of-the-art. NVIDIA submitted a U.S. patent for my invention, and I presented this work at an internal main event held by the Vice President of engineering at NVIDIA. We also wrote a scientific paper on this method, and it is available here.


John Deere Machine Learning Team (May 2020 - May 2022)

  • I developed a fully automated and visual-based local navigation and docking method for the John Deere robot mower. I employed the developed method on a larger scale and as a parking assist for the Fairway mower. I used deep learning object detection followed by a deep Q-network (reinforcement learning agent) serially. This innovation was selected as the top 3 technical innovation at UIUC research park (among about 840 interns in the summer of 2020). We wrote a conference paper on this method, and it is available here.

  • I utilized vision autoencoders for noise reduction in planting conditions.

  • I developed a visual-based precision planting method using deep learning for seed depth and trench profile measurement.

  • I developed an optimized mowing assist using a combination of deep learning semantic segmentation and scene reconstruction algorithms.