Responsible for developing model and inference on resource constrained platforms like Tensilica DSP, ARM and RISCV cores.
Responsible for optimizing and improving algorithm performance in real-world conditions - demonstrating innovative solutions to tough challenges.
Work with cross-functional product team to deliver seamless customer audio experience.
Key Qualifications:
For consideration, you must bring the following minimum skills and experiences to our team:
Experience leading a ML team with 10+ years of experience working in audio signal processing/ML.
Tiny ML / Embedded ML - Hands-on experience porting neural network algorithms from intermediate representations such as Tensor Flow (TFLM), ONNX, etc. onto embedded targets using device-specific compilation tools and/or inference API s.
Deep understanding of on-device quantization techniques including post-training quantization, training-aware quantization, mixed precision inference.
Strong programming skills in c, python.
Conceptual understanding of how neural network operators map to embedded hardware accelerators such as DSP s and NPU s.
Familiarity with Deep Learning Audio Signal Processing approaches for tasks including Speech enhancement / noise suppression / voice pickup
Additional Skills:
Experienced with Linux, Docker.
Familiarity with CMSIS NN, HIFI NNLib is a plus
Familiarity with audio measurements and standard subjective/objective audio evaluation metrics.
Experience working in hardware product teams from product concept to mass production
Good Audio listening skills and experience detecting audio artifacts.
Experience communicating effectively in a cross functional environment.
Strong problem-solving, critical-thinking skills
Familiarity with code version control practices
