Key Responsibility:
1.Production-Grade Algorithm Development:
• Design and implement perception algorithms (based on Vision/Ultrasonic fusion) for low speed scenarios, including obstacle classification (static/dynamic), and ground marking recognition.
• Develop high-precision relative localization solutions using multi-sensor fusion (Wheel Odometry, IMU, Steering Angle) based on EKF/UKF or other filtering frameworks.
2.Embedded Deployment & Optimization:
• Lead the migration of algorithms to embedded platform, e.g. Nvidia/Qualcomm/TI;.
• Perform deep-level code optimization (C++14/
17) and memory management to minimize CPU/RAM footprint while maximizing throughput.
• Utilize hardware-specific toolchains (TensorRT, TIDL, SNPE) to implement model pruning, INT8 quantization, and custom operator optimization.
3.Data-Driven Iteration & Quality Assurance:
• Establish a data-driven loop to identify and resolve "long-tail" perception issues and localization drift in complex environments (e.g., low light, narrow spaces, non-standard markings).
• Conduct rigorous unit testing, SIL (Software-in-the-Loop), and HIL (Hardware-in-the-Loop) validation to ensure algorithm robustness.
4.Functional Safety & Compliance:
• Ensure all software development adheres to automotive standards such as MISRA C++, ASPICE, and ISO 26262 functional safety requirements.
5.Cross-Functional Collaboration:
• Work closely with System, Hardware, and Validation teams to troubleshoot urgent on-vehicle issues, ensuring zero-blocker delivery for critical project nodes.
Qualifications
1.Master’s or PhD in Computer Science, Robotics, Electrical Engineering, or a related field.
2.At least 3 years of experience in mass production programs for low-speed features
3.Professional Exposure
• Mastery of C++ (OOP, STL, Multi-threading) and Python; experience with embedded Linux or QNX. Solid foundation in Deep Learning (BEV, Transformer, CNN) and classical estimation theory (Kalman Filters).
4.Preferred previous experiences
• Experience in at least one full lifecycle of mass production for low speed features (Start of Production experience).
• Hands-on experience in optimizing low-level kernels (e.g., CUDA, NEON) for automotive SoCs.
• Experience with professional automotive tools like CANoe/CANalyzer and HIL simulation.
5.Functional Skills
• System thinking, root cause analysis, statistical data analysis, clear technical writing and presentation skills.
6.Behavioral Skills
• Self-driven, proactive problem solver, strong communication skills, ability to work cross-functionally in global teams.
7.Special Knowledge and Skills Required
• Knowledge of ISO
26262.
8.Any Others
• Ability to manage multiple tasks and prioritize effectively in a fast-paced environment.
