Bricks, a minimalist, intuitive and tailor-made project management tool design for college students while maintain professional, enhances meeting records and organizational management.● Led 4-members as a backend team leader, completed login-registration pages and deployed it on Google Cloud Platform (GCP). ● Collaborated and communicated with the frontend for integration, and completed API document for functions of 4 web pages.● Planned product backlog and 7 sprints for the entire semester, executed planning, demo, and retrospective meetings.● Developed the login system, authored ERD, normalized databases, assisted in code packaging and deployed.

● Developed 50 data cleaning pipelines using Databricks and PySpark, integrating DataLakeHouse, MongoDB, and AWS Redshift to complete the ETL flow.● Redesigned data ingestion processes using AWS Lambda and Databricks, reducing processing time by 10+ minutes.● Migrated AWS Glue code to Databricks Jobs and implemented coding style optimization, reducing costs by 40%.● Built streaming pipelines using AWS EventBridge, Kinesis, and AWS Glue to capture real-time data for monitoring IoT device energy consumption.● Developed a RAG system using AWS Bedrock and AWS OSS to extract and analysis data from PDFs.

Employing statistical methods and Python, paired trading strategies and backtesting frameworks are constructed, drawing from historical price and volume data or other pertinent target information.● Target selection is conducted using Cointegration.● Trading strategies are optimized utilizing the OU-process.● Backtesting is performed using historical data.

● Planned 21 courses over two semesters, earning a satisfaction rating of 4.5/5.● Deconstructing the business analytics framework (Management Consulting, Data Analysis, Strategy Formulation) to design courses.● Identified suitable lecturers and coordinated course content based on members feedback, serving as the liaison with lecturers.● Served as a mentor, providing Tableau and Python support.

Internship focused on leveraging corporate financial statements to develop a model predicting company carbon emissions. Deploy the model on the website to facilitate regulatory assessments, streamline reviews, and assist SMEs in promptly evaluating review costs.● Achieved 98% accuracy in estimating carbon emissions for both in-sample (86 companies) and out-sample (5 companies) data.● Filtered 86 usable companies from a pool of 400, and processed the data into a suitable dataset for training tree-type model. ● Implemented XGBoost regressor to build a predictive model. ● Deploy the model to the website using Docker and Google Cloud Platform (GCP).