Siyu Xiao
Data Scientist Engineer @Fannie Mae
Signup · Get unlimited contacts
WORK HISTORY
Data Scientist Engineer @Fannie Mae
Plano, TX, US
In the Enterprise Modeling & Analytics team, did statistical modeling for Classic FICO score, VantageScore and FICO 10T Score dataset from FHFA (Federal Housing Finance Agency) with Python in AWS SageMakerenvironment, used linear regression and neural network (MLP) to impute the dataset with more years.• Support team directors to answer and fix the questions from AWS CloudWatch and other teams in L1/L2 capacity.• Designed encryption and decryption algorithm for Fannie Mae loan id, used Git to control version of the code.• Evaluated a borrower’s creditworthiness by Credit Bureau Database (CBD) data from TransUnion, Experian and Equifax, did data mining like calculating correlation coefficient and standard deviation. Did data visualization for CBD score analysis result with Matplotlib and did presentation for the whole team.• Accessed AWS Redshift with boto3 library to get FICO and CBD dataset, building required datasets for the team and saved in parquet file, uploaded the datasets to AWS S3.• Created ETL data pipeline in AWS Glue and Step Functions for FICO score machine learning models, used AWS Lambda to run the models without managing servers.• Designed LLM-based solution for real estate mortgage consultation, used Llama-30B as the foundation model, enhanced with prompt engineering and Retrieval-Augmented Generation (RAG), fine-tuning by LoRA• Used Python Prophet library to analyze time-series loan data to detect Fannie Mae anomaly mortgage. Accelerated the code running speed to 100+ times by optimizing code structure and Python Multiprocessing.• In AWS EMR environment, use PySpark to accelerate billions of rows loan data processing.• In AWS Sagemaker Studio, Drew line plots data visualization with Matplotlib and Seaborn libraries to show the loan data trend like Delinquent Active Count, Loan Portfolio and so on.Environment: Python, SQL, AWS, SageMaker, Jira, Confluence, Pandas, Matplotlab, docx2python, TensorFlow
EDUCATION
Harbin University of Science and Technology
Bachelor's degree, Electrical, Electronics and Communications Engineering
Northeastern University
Master's degree, Data Analytics
ABOUT SIYU XIAO
I am currently working at Fannie Mae as a Data Scientist/Engineer. I have about 8 years of experience as Data Scientist and Software Engineer with technical prowess. I have 8+ years experience with performing Data Analysis with compiling, analyzing, optimizing, validating and developing Machine Learning models including neural network models for solving the business problems. I have worked on projects involved AI (Artificial Intelligence), Information Theory, Deep Learning, Machine Learning, statistical modeling, computer vision, data transformation, performed sentiment analytics and handled large datasets. Programming Skills: Python, R, HTML, CSS, C/C++/C#, Java, Scala, MATLAB, LINGO, Lisp, Git/GitHub Database: MySQL, SQL Server, PostgreSQL, Snowflake, AWS Redshift SQL, MS Access, Hive, Oracle, Neo4j Tools: Microsoft Office(including Word, Excel, PowerPoint, VBA, Project, Visio), Tableau, Power BI, SPSS, RapidMiner, Orange, Talend, Informatica, Palantir, LangChain, Hugging Face
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.