Compiler Frontend Engineer
Normal Computing CorporationResponsibilities
- Bring up frontend and mid-level compiler support for new LLM architectures, including attention variants and MoE routing schemes.
- Extend the compiler’s IR, operations, and lowering passes to represent new architectural primitives faithfully.
- Represent model quantization schemes and numeric formats while preserving model accuracy through lowering.
- Track the LLM architecture landscape and recommend architectures for future bring-up.
- Create test coverage to maintain compiler correctness as it evolves.
- Validate compiler bring-up using simulators and FPGA prototypes for pre-silicon hardware.
- Hand lowered programs to backend compiler engineers and collaborate with LLM serving engineers to take models into serving.
Requirements
- Hands-on experience writing and shipping ML compiler passes.
- Deep familiarity with PyTorch internals, ideally including TorchDynamo, FX, and torch.export.
- Working knowledge of modern LLM architectures, including attention mechanisms, MoE routing, and quantization schemes.
- Experience designing compiler infrastructure in a young codebase and extending established compiler infrastructure.
- Comfort building software for evolving hardware through simulation and bring-up.
- Fluency in Python and familiarity with C++ or Rust.
- Ability to follow relevant research literature closely and publish on behalf of Normal.
- Bonus: experience with IREE, XLA, or TVM.
- Bonus: familiarity with JAX or ONNX.
- Bonus: experience partitioning models across devices using tensor, pipeline, or expert parallelism.
- Bonus: experience with hardware-software co-design or non-standard and in-memory hardware targets.
- Bonus: contributions to open-source compiler or ML infrastructure projects.
Categories
About Normal Computing Corporation
At Normal, we're rewriting AI foundations to advance the frontier of reasoning and reliability in the physical world. We are tackling problems across semiconductors and industrials with a mix of interdisciplinary approaches across the full stack: from probabilistic software infrastructure and algorithms to hardware and physics, enabling AI that can reason and understand its own limits. We understand that our technology is only as powerful as the people behind it. Every employee drives significant impact within our products, often working directly with customers and embedding across our tightly-knit team. Our team members are driven by curiosity and passion for solving some of the most challenging problems in the world of atoms. Normal was founded in 2022 by engineers and scientists that pioneered industry-leading Physics + ML tools for next-gen AI at Google Brain and Google X.