Valerie Taylor: Leading research in high performance computing

Valerie Taylor grew up around circuit boards and soldering irons because her father was an electrical engineer who started his own telecommunications company. As a teenager, Taylor learned how to build circuit boards from schematics from her father. He would bring home boards that needed to be finished and show her how to read the schematic to complete the board. This was the beginning of her interest in computer science.

In high school, she took a course in the programming language Fortran. She enjoyed writing code to perform a given task, such as analyze data or handle something such as an airline reservation system. At that time programming was done via punch cards. Special machines were used to type code that was punched onto cards. The teacher disabled the copy button on the machine, forcing the students to rewrite their entire line of code every time there was an error.

This created difficulty. As Taylor explained, ​“You needed the copy button for easy corrections of code. For example, if you forgot to add a comma with the arguments in a function call, you would encounter an error. It is easier to copy your code until that point, add your comma and copy the rest to avoid making another error.”

For her undergraduate degree in Computer and Electrical Engineering, Taylor went to Purdue University, where they had terminals rather than punch cards. That made coding much easier because students no longer had to rewrite their entire code to correct one mistake.

As coding technology changed, Taylor’s fascination with computer science remained. ​“I just really enjoyed programming,” Taylor said, ​“because you could give a computer instructions to do a complex task and get results quickly. I always found that exciting.” Debugging code is tedious but rewarding when the code works. ​“You debug your code based on your understanding of the execution system,” she explained.

Taylor finished her Ph.D. in Computer Engineering at the University of California, Berkeley in 1991. That same year, she presented a paper at the Supercomputing Conference in Albuquerque, New Mexico. At the conference, she mentioned to some attendees that she was starting as an assistant professor at Northwestern University. She was told she had to meet Rick Stevens, the new Mathematics and Computer Science (MCS) division director at the U.S. Department of Energy’s (DOE) Argonne National Laboratory. Stevens encouraged her to visit Argonne and give a talk, which she did.

After that, Taylor began collaborating with Argonne researchers via a joint appointment, while teaching at Northwestern on Tuesdays and Thursdays. Her research collaborations with Argonne focused on performance analysis and modeling of parallel, scientific applications. This joint appointment continued until she went to Texas A&M in 2002.

She stayed in contact with Argonne. In 2016, when she visited Argonne to give an invited talk, Taylor was encouraged to apply for the position of MCS division director. She interviewed and was offered the position in spring 2017. ​“I’ve been having fun ever since,” Taylor said.

As director, Taylor leads a team of 90 scientists and postdocs and develops strategy for the MCS division. She works collaboratively with leadership within her own division and across the lab.

Taylor is also a researcher with projects of her own. She spends most of the work week in meetings, focused on strategy and current initiatives within Argonne as well as collaborations with other labs, universities and industry. On Fridays, she focuses on research. 

Her favorite projects ​“are always the current ones.” Right now, these include a co-design approach for microelectronics research. This project is focused on novel materials for neuromorphic devices for high energy physics detectors. She is also working on the use of large language models for parallel, scientific code generation.

“When I was at Northwestern University with a joint appointment with Argonne, I was a faculty member having a lot of fun collaborating with researchers at Argonne in high performance computing,” Taylor said. ​“Now I have the unique opportunity to lead a division of exceptional researchers at Argonne who are significantly advancing the field of high performance computing.”

Her work is primarily funded by DOE’s Office of Science for Advanced Scientific Computing Research. She has also received funding from DOE’s Office of Science for Basic Energy Science, DOE’s National Nuclear Security Administration and the U.S. National Science Foundation Directorate for Computer and Information Science and Engineering.