Yasp is building the next-generation AI compiler for training deep learning models, powered by LLM-based code generation and hardware-aware optimization. We’re looking for a passionate **Software Engineer** to help shape the future of AI infrastructure by joining our core compiler team.
In this role, you will collaborate closely with ML engineers, compiler developers, and our LLM code generation team to accelerate model training for diverse hardware targets. You’ll play a key role in developing intelligent compiler components that translate AI models into optimized, low-level code tailored for performance.
What You’ll Do
Collaborate with ML and compiler teams to design, build, and optimize the yasp Compiler for training acceleration.
Work with LLM-based agent systems to automate code generation and kernel optimization.
Design and implement compiler passes, runtime optimizations, and backend targeting for AI workloads.
Analyze performance bottlenecks in training pipelines and apply optimizations at both framework and kernel levels.
What We’re Looking For
Strong programming skills in **Python** and **C/C++**.
Strong understanding of **CUDA**, **LLVM**, **Triton**, or custom codegen frameworks.
Experience with AI frameworks (e.g., **PyTorch**, **TensorFlow**, **ONNX**, **TensorRT**) or compiler stacks.
Solid understanding of **machine learning model internals** and training workflows.
Experience in performance tuning and low-level optimization on **GPU, CPU, or AI accelerators**.
Bachelor’s or Master’s degree in **Computer Science**, **Computer Engineering**, or equivalent practical experience.
Bonus Points
Experience working on or with AI compilers, torch.compile, MLIR, TVM, XLA, or similar.
Familiarity with **LLM agent systems** and their role in program synthesis or model compilation.
Contributions to open-source compiler, AI, or ML infrastructure projects.