Our solutions make cameras smarter by extracting valuable data from high-resolution video streams.
Job Description
Responsibility:
Develop an AI inference runtime library for embedded Linux platforms using C/C++, supporting efficient inference for large models and multi-modal models.
Based on the characteristics of hardware platforms such as Ambarella, we implement AI operators and optimize inference processes to enhance the model's running speed.
Responsible for the deployment, integration, debugging, and performance optimization of AI models on board-side devices.
Participate in the problem identification of underlying systems, including Linux debugging, memory/bandwidth/IO and other system-level performance optimizations.
Participate in the design and development of a distributed large model inference framework.
Collaborate with algorithm and hardware teams to produce technical solutions and development documents.
Requirements:
Proficient in C/C++ development, with solid engineering skills (memory management, pointers, multithreaded programming, etc.).
Familiar with Linux system programming (processes/threads, synchronization mechanisms, network programming, etc.).
Prioritize understanding the algorithms of LLM/VLM large models and common optimization methods.
Candidates with experience in Ray/Transformers/vLLM development are preferred.
Understand at least one framework such as TensorFlow, Onnx, or PyTorch.
Possess excellent abilities in independent analysis and problem-solving, as well as strong communication and teamwork skills.
