Job Description

AI Vision Processors For Edge Applications

Our solutions make cameras smarter by extracting valuable data from high-resolution video streams.

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.