Company Description 

Job DescriptionResponsible for deploying AI models (such as E2E, VLM, BEV, occupancy, NLP, etc.) converted from training frameworks into AI cockpit or autonomous driving products.Collaborate with algorithm team to support algorithm engineers in optimizing models, including model compression, quantization, pruning, distillation, etc., to reduce model size and compute complexity.Collaborate with basic software team to optimize model design based on chip hardware resources and software architecture to ensure effective integration and performance of the algorithm in overall system.Research and evaluate different hardware platforms and software frameworks to select the best technical solution for different computing scenarios.Deploy model execution environments based on QNX/Linux/Android.Solve technical problems during AI model deployment, including toolchains, compilation, integration, and execution.Actively communicate with chip vendors to solve problems.

QualificationsBachelor's degree or above in Computer Science, Software Engineering, Artificial Intelligence, or electronic engineering field, with solid knowledge in computer science.3+ years of experience in AI model deployment in autonomous driving or AI cockpits.Knowledgeable with common deep learning frameworks (such as PyTorch, TensorFlow, etc.), familiar with deep learning knowledge such as CNN and Transformer.Familiar with AI models for autonomous driving, experience in deep learning model deployment and performance optimization is preferred.Familiar with Linux/QNX operating systems, including task scheduling, memory management, etc.Proficient in C++/Python programming languages.Proficient in cross-compilation, integration, development, and debugging of embedded software SDKs.Familiar with heterogeneous computing, with a deep understanding of computing resources such as CPU/DSP/GPU/NPU, development and optimization of efficient communication and synchronization schemes, and identification of chip computing bottlenecks.Experience in deploying models on edge-side AI chips is preferred, such as Qualcomm SA8255/Nvidia Orin/Horizon J6, etc. Familiar with related technology stacks such as QNN/OpenCL/TensorRT/CUDA is preferred.No block on reading English specification and technical documents, good oral English is a plus.Strong ability to learn quickly, strong self-motivation, and enjoy sharing and helping others.