Mimi Celis is a member of the technical staff in the Advanced Systems Division of Silicon Graphics (SGI) at their headquarters in Mountain View, California, where she is responsible for the high performance Scientific and Mathematical Software Library. She is part of the Supercomputing Applications group which consists of about 25 people and has smaller groups dedicated to Chemistry, Oil and Gas, Operations Research, Computational Fluid Dynamics and Structure Analysis Applications, as well as producing the Scientific Software library. Most members of the group have Ph.D.'s in engineering, although there are a couple of Ph.D's in computer science and applied mathematics. The main objective of the group is to provide high performance software for the high end multiple processor computers produced by SGI.
"As a group," she says, "we identify the routines that need to be in the library, design and implement the algorithms that will yield high performance on SGI's computers. The library routines are written to be efficient on the range of SGI computers, from the single processor to the multiple processor systems. We help others in the applications group incorporate these routines into their codes, and use their feedback to identify new functionality for the library."
"Most of the applications we work with are for scientific computing. They are used to compute solutions to a variety of mathematical problems. My knowledge in numerical analysis is used to design and develop the methods for computing the solutions to these problems. In the last six months I have been redesigning our linear algebra algorithms and implementing them for SGI's Scientific Library on the new parallel computers. Putting together the Scientific Library has been my main project for the past two years."
Mimi has a B.S. in mathematics from the Universidad Nacional Autonoma de Mexico, and an M.S. and Ph.D. in mathematical sciences from Rice University. Her first job after graduation in 1985 was at IMSL as an algorithms designer for linear algebra codes. She then worked at Ardent Computer (which became Stardent Computer) as a member of a benchmarking group. She joined SGI in 1991 as part of an applications group working on applications for the then new parallel workstations. As SGI became a larger player in the supercomputing market, she joined the newly formed Supercomputing Applications group.
"I always knew I wanted to work in industry," she recalls. "I knew I wanted to work as a member of a team solving real world problems and applying my computer skills."
To develop a background for her type of position, Mimi recommends courses in numerical analysis and algorithms. "Computer science courses are also very useful," she adds. "To get the best performance out of a computer you need to understand the hardware, the compilers and the operating system on the machines. Computing skills are a must for working in industry"
"Look for contacts in industry or people who have contacts in industry. The ability to understand problems in different areas and the ability to communicate with people working in different areas is also very important, it is very likely you will work in a team with people with diverse backgrounds."
Return to Archived Profiles and Forums
