Software Engineer II: AI Compiler Engineer

Job Title

Software Engineer II: AI Compiler Engineer

Role Summary

Develop and optimize a deep-learning graph compiler and runtime that converts neural network descriptions from frameworks (PyTorch, TensorFlow, etc.) into efficient code for embedded and special-purpose platforms. Work on compiler middle/end, runtime integration, and performance tuning for DSPs and accelerators.

Experience Level

Mid-level — typically 3–5+ years of relevant software/compiler development experience.

Responsibilities

Deliver compiler and runtime features to map neural networks to constrained embedded processors and accelerators. Key responsibilities include:

• Design and implement passes in a deep-learning compiler stack to lower NN graphs to efficient code.
• Use and extend compiler frameworks such as LLVM and MLIR for IR construction, lowering, and optimizations.
• Implement and optimize neural-network operators and integrate them into a runtime.
• Develop new mapping/optimization algorithms (loop transforms, polyhedral techniques) for diverse processors and accelerators.
• Benchmark and profile end-to-end network performance on DSPs and special-purpose hardware; iterate to improve throughput and latency.
• Design multiprocessor/multicore partitioning and scheduling strategies for embedded systems.
• Create validation and test programs to verify functionality and performance of the CNN SDK.
• Collaborate with hardware designers, partners, and customers to identify acceleration opportunities and standardize APIs.

Requirements

Must-have technical skills and experience:

• Strong C and C++ programming skills with 3–5+ years of professional experience.
• Proven software development on Linux and Windows, including testing, debugging, and release processes.
• Experience with compiler technologies and frameworks (LLVM, MLIR) and IR lowering.
• Practical knowledge of compilation techniques such as loop optimization, polyhedral models, and IR transitions/lowers.

Nice-to-have:

• Production compiler development experience (3+ years preferred).
• Experience with Python and deep-learning frameworks (TensorFlow, Caffe, PyTorch).
• Experience programming or optimizing for embedded platforms (DSPs, DMA engines) and familiarity with deep-learning compilers (Glow, TVM, XLA).
• Familiarity with common neural-net formats (ONNX, NNEF) and a range of network types (classification, detection, segmentation, RNNs).

Education Requirements

Bachelor's degree in Computer Science or Computer Engineering or equivalent practical experience is required. Master's or PhD in a related field is preferred but not required. The posting explicitly allows equivalent experience in place of a degree.

About the Company

Company: Cadence Design Systems

Headquarters: San Jose, California, USA

Cadence Design Systems is a global electronic design automation company that provides software, hardware, and intellectual property for designing advanced semiconductor chips. With over 25 years in the industry, Cadence is known for its innovative technology solutions and has been recognized by Fortune Magazine as one of the 100 Best Companies to Work For. The company is dedicated to solving complex technical challenges in order to enable customers to create revolutionary products and experiences.