职位描述
职位名称:
Intelligent Cabin AI Full-Stack Engineering
公司:
大众汽车（中国）科技有限公司
发布起始日期:
2026/8/3
职位地点:
合肥
职能:
研发
职位描述:
Brief description of the position:
This role involves developing and optimizing LLM applications, building AI Agent systems with multimodal and tool-use capabilities, and deploying AI on edge/embedded devices. It requires strong software engineering skills, hands-on LLM fine-tuning experience, and the ability to integrate AI across cloud and hardware platforms.
This role involves developing and optimizing LLM applications, building AI Agent systems with multimodal and tool-use capabilities, and deploying AI on edge/embedded devices. It requires strong software engineering skills, hands-on LLM fine-tuning experience, and the ability to integrate AI across cloud and hardware platforms.
Key Tasks:
- Define and implement the end-to-cloud collaborative architecture for LLMs/VLMs, and fine-tune open-source models for automotive scenarios using SFT, LoRA, and RLHF. (Main Responsibility)
- Construct high-quality in-vehicle training datasets and automated data pipelines to support continuous model iteration and performance evaluation. (Main Responsibility)
- Design an intelligent cabin agent orchestration layer to enable complex intent decomposition and autonomous cross-application tool use. (Main Responsibility)
- Implement multimodal fusion interaction (voice, vision, gesture, biometrics) to deliver immersive experiences such as gaze-based wake-up, emotion recognition, and proactive contextual responses. (Main Responsibility)
- Build a user profiling and long-term memory system leveraging vector databases, knowledge graphs, and RAG architectures to enhance knowledge grounding and personalized experiences. (Main Responsibility)
- Optimize model inference on automotive SoC platforms through quantization, pruning, and kernel-level tuning to ensure low latency and high energy efficiency. (Main Responsibility)
- Balance cloud costs and edge responsiveness by optimizing API efficiency and system resource utilization. (Main Responsibility)
- Design and implement robust AI guardrail systems to mitigate hallucinations and ensure determinism of vehicle control commands. (Main Responsibility)
- Enforce strict user data privacy protection and comply with automotive-grade AI safety standards. (Main Responsibility)
- Develop scalable AI backend services using Python and Docker containerization, and integrate sensors, actuators, and in-vehicle communication protocols for hardware-software co-development. (Main Responsibility)
Qualification and skills required for the position:
- Education: Bachelor's degree or above in Computer Science, Artificial Intelligence, Robotics, or a closely related field (embedded systems, edge computing, full-stack AI system etc.).
- Working experience: 5+ years experience in software-hardware co-development, with at least 2 years of proven track record in LLM integration (fine-tuning and deploying), AI Agent frameworks, multi-modality AI, AI apps.
- Knowledge, skills and others:
- Proficient in C++/Python; Rust, Embedded C, or RTOS experience is a plus.
- Skilled in edge AI inference (TensorRT/ONNX/TVM) and model optimization (quantization/distillation).
- Hands-on with RAG, fine-tuning, and Agent development (tool use, memory).
- Familiar with SOA, SOME/IP, DDS, and embedded Linux.
- Proven experience deploying large models on resource-constrained edge devices.
- Strong HMI sense and passion for AI-driven user experiences.
- Smart cockpit experience at OEMs/Tier-1 preferred.
- Independent worker with strong communication and AI-hardware integration skills.