Jiayi Zhang

About Jiayi Zhang

Student at The John Hopkins University Beijing Long March Hi-Tech Co, Ltd.(under China Academy of Launch Vehicle Technology)| Beijing, China Position: Senior Software Engineer Ecological Resources (Climate, Water, Land, Ocean) Map System Based on Spatial Computation Formulated EcoStore, a hybrid Key-value and columnar data model, to optimize storage and retrieval of multi-source spatial data, enhancing sequential and random access. Enhanced spatial data decomposition and distributed spatial index mechanisms, and data encoding to elevate storage resource utilization and spatial index quality. Built server-side in-memory R* tree indexing and balanced small file partitioning to boost retrieval efficiency by 30%. Built a unified spatial metadata management model with HBase, accelerating data loading and enhancing scalability. Optimized spatial analysis processing methods for online/offline computational models, utilizing EcoStore and its metadata. Introduced a rapid, iterative grid-based partitioning method for offline batch processing, supporting spatial queries. Enabled HBase local R* tree index loading and spatial analysis processing interfaces, accelerating queries by 10%. Addressed the rapid online generation of vector tiles for spatial vector data in a distributed environment. Enabled any spatial vector data in EcoStore to serve as source data for visualization, supporting dynamic rendering. Temperature and Humidity Prediction System Contributed to developing a machine learning system to accurately predict temperature and humidity. Developed web applications with Django and Vue.js, facilitating efficient data management and visualization. Trained predictive deep learning models, including CNN and various regression models, using TensorFlow/PyTorch. Tailored regression models to industrial needs, reducing loss from 0.215 to 0.135 via strategic feature engineering and prediction adjustments; Utilized MSE and MAE metrics to assess and ensure the model\'s predictive accuracy and stability. Optimized model error from an initial 0.135, employing and selecting Adagrad from various optimization algorithms to achieve an MSE of 0.085, and deployed applications on Windows servers, publishing through IIS. PetroChina Company Ltd.| Beijing, China Position: Software Engineer Intern Optimized Spatial Visualization and Enhanced Data Handling through Advanced Mapping Technologies Strategic Integration and Transition of Geospatial Mapping Technologies for Enhanced Coordinate Accuracy GIS Platform Development

Experience

1. Software Engineer

   Beijing

   Jan 2024 — Present · Beijing, CN

Education

• The Johns Hopkins University

  Master's degree

• The Johns Hopkins University

  Master of Science - MS

  2024