自动驾驶数据挖掘专家_XC

About the Position

We are seeking an expert in data mining for autonomous driving, focusing on perception and planning algorithms in Suzhou, Jiangsu, China.

Responsibilities

1. 数据能力与体系整合搭建

   • Integrate data capabilities and build a comprehensive system.

2. 感知、Planner专项数据挖掘体系建设

   • Specialize in the data mining system for perception and Planner models, focusing on Corner Case mining, classification, root cause analysis, and iterative review methodologies.

3. 统一数据Tag标签体系搭建与治理

   • Lead the establishment and governance of a unified tag system for perception and planning model Corner Cases, including scenario, traffic interaction, model failure, and extreme condition dimensions. Design strategies for automated identification, precise tagging, classification, and clustering of Corner Cases, and establish a mechanism for label quality verification and iterative governance to efficiently screen and accurately sediment various long-tail and extreme failure cases, supporting the optimization and quantitative statistics of model Corner Cases.

4. 高质量数据集与场景库迭代

   • Based on the specialized tag system, focus on the iterative construction of perception and Planner model Corner Case datasets, specifically expanding rare scenarios, extreme weather, complex traffic interactions, irregular obstacles, and extreme driving conditions case libraries and test sets. Complete invalid data filtering, high-quality Corner Case sample purification and balancing processing, continuously supplement model scenario shortcomings, and repair hidden failure issues, significantly improving the generalization ability and robustness of perception and planning models in complex scenarios.

5. 数据驱动算法闭环迭代

   • Deeply integrate with the iterative needs of perception and planning algorithm models, balancing tool capability iteration and business data delivery: On one hand, continuously optimize Corner Case automated mining, intelligent attribution, and batch sample purification tools and platform capabilities, improving the efficiency and accuracy of long-tail failure scenario mining; on the other hand, undertake core data needs for model iteration, responsible for end-to-end Corner Case problem cases, specialized optimization datasets, and data analysis report delivery. Through multiple rounds of data review and version comparison analysis, drive the convergence of model long-tail problems and optimize extreme condition performance, achieving a closed-loop of algorithm iteration.

Requirements

• Deeply engaged in the iteration of autonomous driving perception and planning (Planner) models, proficient in various model failure features, with rich practical experience in mining, clustering, and root cause attribution of perception and planning Corner Cases (long-tail/extreme/hidden failure scenarios);
• Possess experience in building and governing autonomous driving data tag systems and scenario systems, able to unify multi-module data standards and mining specifications;
• Proficient in using Python, SQL, and big data analysis tools, skilled in large-scale multi-sensor data cleaning, scenario clustering, anomaly analysis, and sample selection;

Nice to Have

• Experience in specialized governance of perception/Planner model Corner Cases, building data systems from scratch, NOA mass production data iteration experience, large model automated mining/intelligent tagging implementation experience, related patents, and technical accumulation.

We Offer

• Competitive salary
• Professional growth opportunities
• Collaborative and innovative work environment

About the Company

[Company description if present]