Yu Tian Email and Phone Number

Web Tool for Natural Language Report Detection and Validation- Designed natural language data analyzing workflow specific for text report data and built end-to-end web application with functions of data extraction, data storage, training data preparation, target case detection, searching results validation. - Created a RESTful web service using Java SpringBoot framework, designed and implemented DAO and built object relational mapping with Mysql database using JPA and Hibernate.- Built a collection of adverse event detection algorithms and data manipulation API using Flask, Numpy, NLTK, Scikit-learn, and interact with large scale binary text data stored on Azure Blobstorage using Azure Blob Service REST API. - Designed and implemented responsive front-end using ReactJS, AJAX, Bootstrap, Material-UI coded in JSX6, CSS3. - Deployed entire set of tool and databases on Azure Web Serivice, CosmosDB, Azure SQL database.Cardiac Signal Synchronization Tool- Created user friendly python tool for synchronizing cardiac output data from different sensors with benchmark data. - Implemented functions such as signal synchronization, synchronization evaluation, covariance plot, outlier detection, missing value handling.MAUDE Database Filter- Created python script for filtering complain case reports related to specific medical device usage based on text description from MAUDE database.

-Designed question-answer (QA) system to extract information from relational database based on natural language questions.-Built NER and text classification system using CRF and LSTM-RNN model.-Prepared 10,000 questions dataset by simulating retail investors' behavior using question templates.-Applied Word2Vec algorithm on over 20 GB financial text dataset crawled from internet to improve accuracy of QA system.-Created rule-based parser to transform natural language date expression to formal date time expression.-Deployed QA system on AWS using Docker container and built API for frontend using Flask.