As I gear up for my second year at Northeastern University, I reflect on my incredible journey so far. One of the standout moments was my Machine Learning Engineer Co-op experience at Lendbuzz, where I integrated vLLM into systems, achieving a 175% inference speedup. 🚀
I’m currently immersed in exciting robotics and AI research at:
Before starting my graduate studies at Northeastern University, my academic and professional journey centered around full-stack development while also exploring diverse areas of research and technology. During my undergraduate years, my coursework focused heavily on full-stack development, and I applied these skills to projects at Wenzhou University of Technology, Wenzhou Bopu Institute of Big Data, and MIE DEMO Studio. Notable achievements included public opinion analysis, data-driven research, and developing a volunteer management platform, which was recognized as a national innovation project.
After graduation, I worked as a Technical Manager and Project Leader at Zhejiang Chongxiao Zhong Han Medical Technology Co., where I led full-stack development projects. Key contributions included creating a VC++-based database integration tool and a Springboot+Vue electronic invoice management system, achieving an 80% improvement in operational efficiency.
Alongside my primary focus on full-stack development, I also explored other areas such as sentiment analysis, knowledge graph construction, and user behavior modeling, broadening my technical expertise and research interests.
In the early phase of my graduate studies, I joined the CHATS Lab at Northeastern University, where I worked on chatbot-related projects, contributing to advancements in conversational AI systems.
Feel free to reach out! ✨ Whether for collaboration, brainstorming, or just exchanging ideas, I’d love to connect and explore new possibilities together. 💡
Thank you for visiting my page. Let’s make something great together!
Master in Information System, 2023 - present
Northeastern University
B.Eng. in Computer Science, 2016 - 2020
Wenzhou University of Technology
Keyword: Open World Manipulation With Planning, Grasping Learning, Equivariant Model
Equivariant Models for Robotic Grasping: Developed an advanced robotic grasping model incorporating SE(2) and safe-z prediction, reducing training epochs by 50%.
Vision-Language Models for Strategic Grasping: Integrated LLMs for “Open World Manipulation with Planning,” enhancing robot adaptability in dynamic environments. Developed ThinkGrasp, a vision-language grasping system utilizing GPT-4o, achieving a 98.0% success rate in cluttered scenes, outperforming previous methods by over 50%.
Generative Models for Manipulation Policies: Contributed to the development of IMAGINATION POLICY, a generative point cloud model for key-frame manipulation, improving success rates on RLbench tasks by 20%.
In-service period
AI-Powered Document Insight Engine
PATPAT is an innovative project that aimed to create a versatile chatbot capable of operating across web, iOS, and Edge Devices. Initiated as a vision to overcome the limitations of early LLM technologies, the project combined React, Langchain, Flask, and Swift to build a comprehensive solution for real-time interaction and automation.
A project that implements Deep Q-Networks (DQN) in OpenAI Gym’s CartPole environment, employing equivariant model techniques to achieve faster training speeds.
This project enhances a robotic grasping model using PyTorch by integrating safe-z prediction into an SE(2) model, achieving faster training and optimization for patch-based input.
A comprehensive digital platform aimed at revolutionizing traditional university information systems by centralizing data management and enhancing user experience for students, professors, and administrative staff.
This project delves into optimizing matrix operations using CUDA, demonstrating iterative improvements in addition and multiplication algorithms to enhance performance on GPU architectures.
Conducted Named Entity Recognition (NER) on JD.com product titles in specific categories
An Independent Project leveraging PyTorch to develop a question and answer bot based on a legal-like question and answer dataset using a Seq2Seq model.
An Independent Project focused on developing a Named Entity Recognition system for legal texts using PyTorch with a BiLSTM + CRF architecture.
Converted images back to the underlying chemical structure annotated as InChI text
Collaborative Management System of Chongxiao Zhong Han Medical Technology Co.
Electronic invoice management system based on hospital business
This tool is responsible for setting up the database environment when deploying enterprise SPD software within the hospital.
Recommender system based takeaway platform
A legal platform based on big data and artificial intelligence technology to serve legal practitioners and consultants
Provide a scalable short text similarity calculation scheme
Public opinion analysis based on Wenzhou News Network
Logistics-based recommendation platform for over-the-counter proprietary Chinese medicines
Intelligent online teaching platform