Bosch Rexroth logo

感知模型剪枝量化工程师_XC

Bosch Rexroth·Salary not specified

Nice to have's

PyTorchONNX RuntimePythonC++TensorRT-LLM

Job description

About the Position

We are looking for a Perception Model Pruning and Quantization Engineer to join Bosch Automotive Products (Suzhou) Co., Ltd. This role involves optimizing model deployment on resource-constrained platforms to ensure efficient performance.

Responsibilities

  • Establish a joint optimization pipeline for pruning and quantization models to support efficient deployment of perception models on production platforms with limited computational power. Balance accuracy, latency, throughput, and memory usage.
  • Analyze the computational load and sensitivity of each model module, develop a phased compression strategy, and ensure that the accuracy and robustness of the model do not significantly degrade after compression. Address issues such as operator compatibility and bit consistency.
  • Set up a process for aligning and validating the accuracy of models before and after compression, outputting compression evaluation reports and driving the model iteration loop.

Requirements

  • Bachelor's degree or above in computer science, electronics, automation, mathematics, or related fields. At least 2 years of relevant work experience.
  • Solid foundation in deep learning, familiar with CNN/Transformer architectures, and experience in training or tuning models for object detection, segmentation, or BEV perception.
  • Familiarity with deep learning frameworks such as PyTorch, and inference engines like DPX Platform, TensorRT, ONNX Runtime, and common compression tools.
  • Understanding of model quantization and pruning principles, with experience in INT8 quantization or pruning deployment.
  • Strong engineering skills (Python/C++), familiar with model export, conversion, inference deployment, and performance analysis processes.
  • Preferred: Experience in autonomous driving/ADAS perception or embedded platform deployment.

Nice to Have

  • Experience with structured pruning, sparse training, knowledge distillation, or low-bit quantization for model compression.
  • Experience with NPU/BPU/GPU/embedded platform performance optimization.
  • Experience in delivering in-car production projects.

We Offer

  • Competitive salary and benefits
  • Opportunity to work in a leading automotive technology company
  • Professional growth and development opportunities

About the Company

Bosch Automotive Products (Suzhou) Co., Ltd. is a subsidiary of Bosch Group, a global leader in automotive technology and solutions. We are committed to innovation and excellence in automotive electronics and systems.

© Bosch Rexroth. This job description was sourced from the employer's public career page. TheJob is not the employer — we index the posting and route candidates to the source. All content rights and hiring decisions belong to the employer.

Popular searches