“A lot of students begin a data science program with dreams of working for a big tech company, but there are many more DS jobs out there,” said Rice University Master’s of Data Science (MDS) alumna Xinyang Song ’22, who works as a yield analysis engineer for one of the world’s largest semiconductor manufacturers.
“My undergraduate degree was in Automation, which gave me opportunities to learn basic electrical engineering and computer science knowledge. Until I began my data scientist job search, it never occurred to me to leverage that background. Now I know that electrical engineers with data science degrees can tap into a unique job market.”
Yield analysis needs data science training
Song works closely with process integration engineers, who manage the manufacturing process of semiconductor wafers. As a yield analysis Engineer, her coding skills are used in data wrangling and data analysis, as well as statistical modeling. She interprets the results, then produces and explains visualizations to audiences who are more focused on the interpretation compared to the details.
“Although I was drawn to Rice’s MDS program for its machine learning courses, my job as yield analyst focuses more on the statistics part of my training. Regardless of the tools I use, I still solve each problem with logical thinking and data-driven perspectives,” said Song.
Specializing in machine learning
She first discovered machine learning as an undergraduate. She chose Rice’s MDS program because it covered all the basics of data science like statistics, programming, and visualization and offered a specialization in machine learning.
“Excellent professors taught me machine learning, deep learning, and how to use these AI tools to build projects,” Song said. “The entire program gave me confidence in myself as a data scientist. I felt I knew every step of the data science cycle, from collecting and cleaning the data to building the model and communicating the results through visualizations.”
Storytelling with data
With each project, her ability to tell its story improved. When she presented her capstone project --creating a model to search for key terms in a global law firm’s contracts-- to the client, Song felt comfortable identifying and explaining only the aspects that mattered to the client. “ It was very important to talk to them about how our tool could help them in their daily activities—the inputs and outputs of our model. They did not care about the programming details.
“That was a very good experience for me,” said Song. “Not all graduate students have a real-world solution they can build and talk about. When we were interviewing for internships and new-hire roles, it was obvious the recruiters expected us to be able to discuss those kinds of experiences.
“Working with the new students as a teaching assistant (TA) in the Python course also strengthened my technical and communication skills. As a student, I only needed to know one of the best ways to solve a problem. As a TA, I had to help students who chose different approaches. The students’ questions forced me to think more about the solutions they were considering and the small differences that might make one a slightly better choice. Learning additional solutions and details made me a stronger programmer.”
Working as an analyst
Song now collaborates with fellow engineers, engaging in technical discussions to identify and address systematic issues in the manufacturing process. Her role encompasses three primary responsibilities involving data organization, in-house data analysis methods, and data evaluation, focusing on statistical analysis to communicate meaningful insights.
Now Song talks about technical ideas with other engineers as they identify causes of systematic issues during the manufacturing process. As such, her role includes three primary job duties.
“First, we organize data to improve and refine the data, getting it ready for analysis. Second, we perform data analysis using methods built in-house for our proprietary systems. Finally, we evaluate the data: processes change over time, and we evaluate the percentage a proposed change may have based on our statistical analysis methods. Then we tell its story.”
To learn more about program requirements and coursework, visit the Department of Computer Science.