Electrical and Computer Engineering - Machine Learning and Data Science
GPA: 3.75/4.00
September 2018 - PresentElectrical Engineering and Computer Science
Rank: 3/28 GPA: 4.05/4.30
September 2013 - June 2017Achieved 3 kinds of dashboards on Splunk for monitoring the CPU, memory usage, local storage usage, the health of thread, etc. in each endpoint.
Developed automatic alert system via slack using anomaly detection in Splunk Machine Learning toolkit.
Expedited problem-solving time about 30% by my Splunk App.
Investigated autonomous driving semantic segmentation model using RGB and Depth information..
Achieved 71.3% mIoU on the Cityscapes dataset with 3.3% improvement over baseline.
Improved model performance by 11% by conducting data cleaning and data engineering, imputing missing values.
Achieved top 11% over 1100+ competitors using XGBoost.
Developed RESTful API through Express.js with searching engine, adding new restaurant and edit restaurant information features.
Linked application with a NoSQL database designed by myself using MongoDB.
Developed responsive web design using CSS Bootstrap.
Developed server side application components through Express.js with member register/login/make order/edit order/order confimation email service.
Designed MySQL database schema with member, product and order features.
Utilized Postman to fully examine functionalities.
Implemented Model View Control architecture using Django web framework with login/register, shopping cart, order checkout features.
Developed responsive web design using CSS Bootstrap.
Improved about 6% accuracy by implementing collaborative filtering of user and book similarities.
Formulated the task to pairwise preference prediction solved by latent factor model (SVD algorithm).
Achieved top 21%/ 28% among 423/ 847 competitors on Rating/ Read Prediction.
Proposed and Established meeting speaker straws drawer system with adding new member feature/ admin-login feature.
Designed SQL database schema and linked with Flask REST API.
Enhanced code readability and simplified functionality augmentation using Jinja2 template engine.
Addressed objects missing issue in captions by implementing CNN-LSTM networks and attention mechanism.
Achieved BLEU-4 score of 0.1968 on the MS COCO dataset.
Incorporated a domain classifier into two-stream FCN8s to align cross-domain features.
Achieved 5-10 % mIoU improvement in the SYNTHIA dataset.
Built partical filter to fuse and filter data from odometry, IMU and LIDAR scans to localize household robot.
Projected RGBD information from Kinect to build 2D map using calibration parameters.
Built an image quality assessment model to eliminate poor training data enhancing detection accuracy.
Utilized template matching and k-means to detect optic disc and macula.
Designed a user interface that combined detection and disease classification functions.
Quisque eget nisl id nulla sagittis auctor quis id. Aliquam quis vehicula enim, non aliquam risus.
Coordinated a 6-day, 120-person summer camp event; supervised preparation.