MLIR Compiler Lead (m/f/d)

  • Dresden

Webseite SpiNNcloud Systems GmbH

Unique solution combining Deep Learning, Symbolic AI, and Neuromorphic computing Our platform provides a real-time, low-latency, and energy-efficient cognitive AI platform leveraging cutting-edge research from the Human Brain Project

As a deep-tech startup, we are looking for talented and passionate people with an appetite for problem solving. Besides the minimal requirements for this job position, your profile is a good fit to our company if you have the following values:

  1. High flexibility and adaptability.
  2. Tranquility to work under pressure.
  3. Appetite for learning and problem solving.
  4. Critical thinking.
  5. Ability to communicate effectively.
  6. Keen to collaboration.
  7. Proper time management.
  8. Highly autodidact, independent and proactive.

Responsibilities:

  • Lead the development of the compiler that partitions and maps large-scale models (e.g., including but not limited to the trendy LLM) to the world’s largest brain-like supercomputer, including both processing and memory considerations (on- and off-chip).
  • Contribute to the design of the dialect for our higher and intermediate representation layers.
  • Contribute to the design of the low-level scheduling strategies of parallel tasks onto the processors, data movements and inter processor synchronization.
  • Participate in the design of a multi-pass compiler architecture involving graph optimization passes, data reshaping, constant folding, tiling, padding, among other optimizations to deploy constrained operations in our distributed architecture.
  • Coordinate cross-functional and diverse teams that involve high-level application developers, low-level experts, hardware designers, and PCB engineers.

Minimum Requirements:

  • Experience leading, growing, and nurturing agile development teams while managing agile project frameworks.
  • Experience with CI/CD flows, code review, bug triage meetings and implementing agile testing practices.
  • Implement best practices to monitor progress and leverage dependencies with cross-functional teams to satisfy the diverse needs of numerous stakeholders.
  • Strong knowledge of Multi-Level Intermediate Representations (IR) ranging from low-level backend-specific descriptions, tensor-based descriptions, and high-level dialects.
  • Strong programming skills and development experience in Python, C or C++, involving data structures, graphs, and object-oriented programming.
  • Strong soft skills such as leadership, engagement, self-motivation, and goal-orientation.
  • BSc, MSc or Dipl.-Ing. in Computer Science, Electrical Engineering, Computer Engineering, or any other related computationally intensive field.

Added value:

  • Experience building or maintaining MLIR dialects.
  • Experience porting backends into open-source machine learning compiler frameworks such as Apache TVM or TensorFlow XLA.
  • Experience with the LLVM compiler and toolchain.
  • Strong knowledge of the Tensor Operator Set Architecture (TOSA) or the Tensor Intermediate Representation (TIR).
  • Low-level hardware knowledge.
  • Experience architecting, leading, or developing complex software projects.

We offer a highly competitive salary with reallocation benefits in a flexible and inclusive work environment. We are an equal opportunity employer, and hence we welcome people of different backgrounds, nationalities, and experiences.

Job Category: Software

Job Type: Full Time

Job Location: Dresden

Your contact: Hector Andres Gonzalez Diaz

Location: SpiNNcloud Systems GmbH, Freiberger Straße 37, 01067 Dresden

Um dich für diesen Job zu bewerben, besuche bitte spinncloud.career.softgarden.de.