Embedded ML Inference & Optimization Engineer

Applied Intuition Inc.; Sunnyvale (CA); Hybrid; USD 159,053 - 199,295; Full time

Job summary

Applied Intuition Inc. is seeking a software engineer with deep expertise in optimizing ML models for production-grade embedded runtime environments.

You will work across the ML framework stack and target a range of embedded compute platforms used in on- and off-road ADAS/AD stacks. You will collaborate with ML engineers and software developers, profiling model performance, implementing pruning/quantization, and driving efficient architectures for memory-constrained devices.

Qualifications

• Bachelors in Electrical Engineering or Computer Science, or related field.
• 3+ years of experience with ML accelerators, GPU/CPU/SoC architecture.
• Strong software development skills with embedded programming focus.
• Experience profiling and optimizing model performance on embedded compute platforms.
• Experience working with deep learning frameworks (PyTorch, JAX, ONNX).

Responsibilities

• Drive ML performance optimization across technologies for embedded ADAS/AD stacks.
• Develop compute usage strategies to optimize inference latency and efficiency.
• Work on model pruning and quantization for memory-constrained platforms.
• Collaborate with ML engineers and software developers to optimize model architectures.
• Set up profiling methodologies to identify performance bottlenecks on target hardware.

Skills

ML frameworks experience

Embedded programming

Model profiling

DL frameworks (PyTorch)

Education

Bachelors in Electrical Engineering

Bachelors in Computer Science

Job description

Applied Intuition Inc. is seeking a software engineer with deep expertise in optimizing ML models for production-grade embedded runtime environments.

You will work across the ML framework stack and target a range of embedded compute platforms used in on- and off-road ADAS/AD stacks. You will collaborate with ML engineers and software developers, profiling model performance, implementing pruning/quantization, and driving efficient architectures for memory-constrained devices.