Hybrid Data Scientist, Perception, ADAS Auto-Labeling

at Woven Planet

• Job title: Data Scientist, Perception, ADAS Auto-Labeling
• Job type: Full Time
• Experience level: Junior, Mid level
• Salary: Not specified
• Degree requirement: Bachelor's Degree
• Tech skills: AWS, Azure, Cloud, Google Cloud Platform, Numpy, Pandas, Python, Spark
• Location requirements: Hybrid, Tokyo, Japan

About the role

• Data Scientist focusing on perception technologies and automated driving solutions for Woven by Toyota. Collaborating with cross-functional teams and utilizing large-scale data and machine learning techniques.

Responsibilities

• Come up with data strategies on how to better leverage our labeled datasets to improve the performance of our auto-labeling pipeline (data sampling)
• Develop and maintain data pipelines to monitor annotation quality metrics, model failure patterns, and dataset characteristics at scale using data warehouses and distributed processing tools
• Collaborate with the ML team to diagnose model bottlenecks through statistical analysis and data slicing, identifying whether issues come from insufficient training samples, distribution shifts, or systematic biases in the annotation process
• Participate in data-related activities within the team, across the company in collaboration with Toyota group, and enhancing team members' capability for data science

Requirements

• 2+ years experience in following industries or research areas:
• ・AD/ADAS
• ・Robotics
• ・Computer Vision
• 2+ years experience in data science or related areas, including theoretical aspects of data science like machine learning (deep learning, statistical analysis, and mathematical modeling)
• Experience writing software using 1) Python for data science (numpy, scipy, scikit, pandas), 2) database, and 3) cloud platform services (AWS, GCP, Azure)
• Bachelor's degree in science or engineering
• Business-level proficiency in English
• NICE TO HAVES
• Master's degree or Ph.D. in related field
• 5+ years experience in data science or related areas
• Hands-on experience in the following:
• ・Experience with computer vision datasets and annotation pipelines, particularly for autonomous systems or multi-camera setups, with a track record of identifying and resolving data quality issues
• ・Familiarity with active learning strategies and uncertainty quantification techniques to prioritize which samples need human review or re-annotation
• ・Proficiency with data visualization tools and statistical methods for large-scale dataset analysis, enabling quick identification of distribution shifts, labeling inconsistencies, or underrepresented scenarios in the training data
• ・Strong Python programming skills with experience using Git for version control and collaborative development in a team environment
• ・Hands-on experience with data infrastructure tools such as data warehouses, dbt for data transformation, and Spark for large-scale data processing
• Business-level proficiency in Japanese (especially, smooth reading and listening)

Benefits

• Competitive Salary - Based on experience
• Work Hours - Flexible working time
• Paid Holiday - 20 days per year (prorated)
• Sick Leave - 6 days per year (prorated)
• Holiday - Sat & Sun, Japanese National Holidays, and other days defined by our company
• Japanese Social Insurance - Health Insurance, Pension, Workers’ Comp, and Unemployment Insurance, Long-term care insurance
• Housing Allowance
• Retirement Benefits
• Rental Cars Support
• In-house Training Program (software study/language study)