Anna Verner
About Anna Verner
Anna Verner is a Member of Technical Staff at Cerebras Systems, where she has worked since 2022. She specializes in designing and implementing performance-optimized matrix multiplication kernels and has a background in software engineering from Intel Corporation.
Work at Cerebras Systems
Anna Verner has been a Member of Technical Staff at Cerebras Systems since 2022. In her role, she has designed and implemented standalone multi-variant matrix multiplication kernels to support a new composite attention kernel. This work has unlocked multiple hardware-specific performance and memory optimizations. Additionally, she proposed API changes that facilitated the merging of several broadcast kernel modules, which reduced code complexity and maintenance efforts. Her contributions also include implementing features such as 3D-tensor support and multi-lane support for softmax/matmul, enhancing the capabilities of the composite attention kernel.
Previous Experience at Intel Corporation
Prior to her current position, Anna Verner worked at Intel Corporation. She served as a Software Engineer from 2019 to 2020, where she contributed to various software engineering projects. Additionally, she completed two summer internships at Intel in 2017 and 2018, each lasting one month. During these internships, she gained practical experience in software development and engineering processes.
Education and Expertise
Anna Verner studied Applied Mathematics and Computer Science at Northern (Arctic) Federal University. She achieved both a Specialist's degree and a Master's degree in this field. Her educational background provides her with a strong foundation in mathematical principles and computational techniques, which she applies in her professional work.
Technical Contributions and Innovations
In her role at Cerebras Systems, Anna Verner has made significant technical contributions. She proposed and implemented a kernel API extension to support multi-variant kernels, enhancing the functionality of Cerebras' AI chip. Furthermore, she has worked closely with stack and runtime engineers to address tasks and issues related to kernel library and runtime interaction, which has improved system integration.