Compiler Engineer

Senior / Principal Compiler Engineer
Bristol preferred | Fully remote within the UK considered Exceptional candidates elsewhere in Europe may also be considered Photonic computing | C++ | MLIR | LLVM
I’m supporting a well-funded UK deep-tech company developing a proprietary photonic computing platform for high-performance, privacy-preserving workloads. They’re looking for a Senior or Principal Compiler Engineer to help build a new compiler stack targeting their novel hardware architecture.

You’ll work across compilers, machine-learning frameworks and hardware, translating high-level workloads into efficient code for a proprietary execution target.

What you’ll be doing

Designing and implementing core components of a greenfield compiler stack
Developing intermediate representations, lowering pipelines and transformation passes
Building hardware-aware compiler optimisations
Integrating with technologies such as Google HEIR, PyTorch and ONNX Runtime
Translating high-level workloads into efficient code for a novel accelerator
Working closely with hardware, simulation, cryptography and algorithms engineers
Benchmarking, profiling and debugging generated workloads
Influencing the technical direction and architecture of a growing software platform
What they’re looking for:
Strong professional experience with modern C++
Hands-on compiler development experience at the IR, middle-end or back-end level
Experience with MLIR, LLVM IR or another compiler/framework intermediate representation
Knowledge of compiler passes, lowering, optimisation and code generation
Experience targeting a processor, accelerator or custom architecture
An interest in hardware-software co-design and building production-quality systems
MLIR experience would be ideal, but it isn’t essential.

Strong LLVM back-end or intermediate-representation engineers who are interested in learning MLIR are encouraged to apply. Experience with GPU, NPU, DSP, FPGA or custom-silicon compilers would be particularly relevant. Previous cryptography or Fully Homomorphic Encryption experience is useful but not required. This is an opportunity to build a compiler around genuinely novel hardware and have significant influence over the software architecture at an early stage. Please message Harry Hansford if you’d like further information or a confidential discussion.