Compilers for neuromorphic computing


Neuromorphic computing denotes brain-inspired systems for artificial intelligence (AI). It promises vast performance improvements and power savings compared to conventional von-Neuman architectures. These systems are researched at RWTH within multiple large-scale projects.

In this thesis, you will participate in our compiler team to develop LLVM-MLIR infrastracture to deploy any AI workload to neuromorphic architectures.


LLVM is one of the standard compilers. Recently, MLIR extended it to address the needs of AI workloads. At ICE, we develop the first-ever fully-integrated compiler for neuromorphic systems. You will be able to participate in a novel, evolving, and relevant open-source community that large companies and major academic institutions drive.


  • MLIR frontend development linking AI frameworks like Tensorflow to our neuromorphic architecture
  • LLVM backend for core programming
  • Verification in simulation and with RTL protoypes
  • Architecture optimizations



  • Basic C++ knowledge
  • Interest in compiler topics
  • Willigness to work with large-scale open-source projects

Knowledge in RTL, Python, ML is a plus.