Auto-grader for ML class blind test casesDeveloping an advanced auto-grader system for evaluating student ML algorithm code in a remote team setting.Collaborating with team members across different locations, utilizing effective communication and version control systems; designing and implementing the auto-grader in Python, leveraging its libraries and frameworks for machine learning and data analysis.Implementing blind testing methodology, using sequestered truth data to generate decision statistics from student code.Collaborating with instructors and teammates to establish evaluation metrics and generate ROC curves; ensuring accuracy and reliability through rigorous testing and debugging processes.Optimizing the auto-grader for large datasets, utilizing efficient data handling techniques.Conducting code reviews and providing feedback to improve the system's quality and efficiency; documenting the design, implementation details, and operational procedures for ease of use and future maintenance.

• Developed user interfaces (UI) using Visual Studio Code to facilitate trading operations at multiple banks, while utilizing NetBeans for server management and database synchronization.• Implemented a Java, SQL, and iReport-based service within the UI to automatically generate transaction record reports, streamlining reporting processes; enhanced client experience by populating a new stock options filter parameter in the database, enabling effective data filtering for clients; created a C# filter to optimize data filtering capabilities for interface users, improving efficiency and usability.• Ensured smooth operation of the financial trading system by conducting thorough testing, bug fixing, and maintenance tasks, resulting in 99% system reliability and 100% client satisfaction.

Academic publication: Xingyu Yang, Dong Qiang, Zixuan Chen, Hao Wang, Zhengxue Zhou, and Xuping Zhan. ”Dynamic Modeling and Digital Twin of a Harmonic Drive Based Collaborative Robot Joint.” (Under Review, IEEE International Conference on Robotics and Automation, 2022.)• Developed a sophisticated dynamic model for a single collaborative robot joint, focusing on the interplay among its components, to optimize its performance and responsiveness.• Experimented with machine learning algorithms in Matlab, with the aim of identifying and calibrating system parameters of the dynamic model to mitigate unstructured uncertainty.• Prepared and presented an in-depth report to the supervisor and team members, outlining the project objectives, methodologies used, and outcomes, fostering an enhanced understanding of the project; engaged in collaboration with a multidisciplinary team, providing insights that led to a more robust and efficient robot joint design.

- Conducted in-depth research on the basic principles of Non-Orthogonal Multiple Access (NOMA) and gained a comprehensive understanding of its applications and benefits in wireless communication systems.Performed a literature review to gather relevant information on NOMA, including its technical foundations, key concepts, and potential advantages over traditional orthogonal multiple access schemes; explored different scheduling algorithms used in NOMA systems, examining their strengths, weaknesses, and suitability for various network scenarios.- Utilized MATLAB to implement simulation programs and test the performance of three different scheduling algorithms in a NOMA context.Designed and developed MATLAB programs to model the behavior of the NOMA system and simulate the performance of the selected scheduling algorithms; generated test cases to evaluate and compare the efficiency and effectiveness of the scheduling algorithms, considering metrics such as throughput, latency, and fairness; conducted comprehensive simulations, collecting and analyzing data to assess the performance trade-offs and advantages of each scheduling algorithm in terms of system capacity and resource allocation.- Documented the research findings, simulation methodologies, and results in a detailed report or technical document.Summarized the key insights and observations from the literature review, highlighting the main contributions and advancements in the NOMA field; described the MATLAB simulation setup, including the program structure, input parameters, and simulation environment; presented the results of the simulations, including performance metrics and comparative analysis of the three scheduling algorithms; provided an objective evaluation and interpretation of the results, discussing the implications for future research and practical implementation of NOMA systems.