Jinli Duan

Relevant Coursework: Machine Learning, Deep learning, Internet Architecture etc.

Relevant Coursework: Python, Database, C++, Java, Linear Algebra etc. Former vice president of Astronomy club at UCI.

AP physics C, AP Calculus AB/BC.

Volunteered as a Teaching Assistant for a local elementary school for a duration of six months. Assisted over 100 students in STEM classes. Provided academic support and guidance to enhance students’ understanding of STEM concepts.

Engaged in the development of Zopeak Star, a building information modelling software tailored for curtain wall design. Facilitated effective communication and collaboration between the front-end and back-end development teams. Methodically gathered and analyzed stakeholder requirements, subsequently composing comprehensive requirement documents. Under the expert supervision of Dr. Sun, the Chief Executive Officer.

Actively participated in the management and documentation of web development tasks. Undertook responsibilities in maintaining the Glodon web system, ensuring its seamless operation and performance. Made significant contributions to backend support in a 3D modeling project that leveraged AutoCAD technology.

Provided support to Professor Gassko as a Learning Assistant for ICS6B, an intensive discrete math course, throughout a complete academic quarter. Convened comprehensive exam review sessions, held regular office hours, and actively participated in lectures. Skilfully facilitated exam preparation and elevated comprehension of course material through personalized and tailored guidance and instruction.

Developed an online application for 3D Visualization and Holographic Display of Scatter Plots. Utilized three.js for advanced rendering. Enabled users to save and load completed 3D modeling projects from the server. Prepared comprehensive persona and user guide documents to optimize user experience.

Developed a cutting-edge application for Patient Voice Recording Analysis and Segmentation. Leveraged machine learning and data visualization techniques. Demonstrated the application’s efficacy at the esteemed National Capstone Design Conference.

Preprocessed datasets and extracted textual features using TF-IDF. Implemented the FNN neural network algorithm for efficient search and tailored diet recommendations. Benchmarked with KNN/SVM.

Successfully addressed the challenge of long baseline matching in perspective RGB images with limited overlap. Inferred spatial connections and established corridors between distantly related, partially overlapping images. Conducted a thorough application and comparative analysis of both supervised and unsupervised learning algorithms, including VGG16, ViT_B16_224, and SuperGlue.

Utilized key prefix for efficient in-memory storage. Employed a hash mechanism to map key-value pairs to a disk-based slice. Implemented buffering to handle insert bursts effectively. Benchmark against various data structures including B+Tree, SA B+Tree, Kraska B+Tree, and others for performance evaluation.

Preprocessed job listings to create a structured dataset. Employed BERT model on the job listing dataset to convert it into numpy arrays and calculated similarity scores using cosine similarity and KNN+KDTree. Developed a personalized recommender system based on user profiles and search history.

Designed and implemented a comprehensive system to enhance the robustness of Scientific Workflows. Successfully replicated workflows using reprozip and efficiently managed them with Kubernetes. Streamlined the debugging process using Bugdoc and automated the upgrade process using VersionClimber.