Abstract:
Performance is the currency of modern computing. Achieving peak throughput on 
fast-evolving accelerators demands more control than today's compilers provide. I will 
present Exo and the Exocompilation paradigm: a user-schedulable programming language 
that shifts two responsibilities traditionally hard-coded in compilers--hardware backends
and optimization strategies--into safe, extensible user libraries. In Exo, optimizations are 
expressed as verified rewrites that guarantee functional equivalence, and low-level 
primitives (e.g., explicit instruction selection, GPU asynchrony) give programmers precise 
control. This design enables targeting CPUs, GPUs, and matrix engines, and supports 
concise, reusable scheduling libraries that match or surpass the performance of state-ofthe-art libraries such as cuBLAS and MKL. Exo-generated kernels already ship at scale, including on Apple devices.

Bio:
Yuka Ikarashi is a sixth-year PhD candidate at MIT CSAIL. She received her MS from MIT in 
2022 and her BS from the University of Tokyo in 2020. Passionate about compilers and 
programming languages for high-performance computing, she created the Exo 
programming language. She has previously worked at Apple, Amazon, and CERN, applying 
her research to various accelerators and applications. She has been awarded the Quad 
Fellowship, the Masason Foundation Fellowship, the Funai Foundation Fellowship, the ML 
and Systems Rising Stars Award, and the Rising Stars in EECS Award. Originally from 
Tokyo, Japan, she is an avid traveler and has visited 36 countries.