AI Quality evaluator -Japanese Content (12-month contract)

Location: Tokyo, Tokyo, jp, Remote

Company Description

Canva's Localisation team sits within the International Supergroup and is responsible for bringing Canva's product and AI experiences to users in their language, across the world. With international users representing almost 60% of Canva's user base, localisation is a core strategic priority and ensuring our AI features meet the quality bar for LOTE markets is one of our biggest opportunities.

We partner closely with our AI Research and AI Platform teams to deliver in-language AI evaluations, dataset creation, and quality assurance across Canva's pioneer markets. As Canva expands its AI capabilities globally, we are building dedicated in-language resourcing to support this growing body of work.

Job Description

We are looking for a labeller with native proficiency in one of Canva's priority languages combined with a strong eye for visual communication and/or experience in language quality evaluation. The ideal candidate thrives in a fast-paced, high-energy environment and can handle a variety of tasks — from repetitive manual labelling to in-language quality evaluation of AI-generated content. The candidate should be good at managing their own time and capacity across a varying number of concurrent projects. They should be adaptable and have strong attention to detail in order to work with the different Canva labelling guidelines and standards.

Language proficiency required in Japanese.

Qualifications

Support international AI company goals by handling labelling and annotation tasks in your Japanese, in collaboration with the Data Labelling Teams based in PH and EU. This includes delivering high-quality, accurate, and timely labelling outputs that directly support ongoing product, content, and AI development initiatives.

0-12 month deliverables

• Complete assigned labelling tasks from internal customer requests in alignment with project priorities.

• Maintain accuracy, consistency, and efficiency across all labelling outputs.

• Meet delivery timelines and output targets for each task type.

• Actively support the Data Labelling team’s objectives through collaboration, responsiveness, and adherence to Canva’s quality and data standards.

• Complete a variety of labelling tasks based on provided guidelines within a limited timeframe in order to support Canva’s AI feature training and development.

You’ll do this by:

• Labelling, annotating, and evaluating photos, graphics, videos, stickers, designs, and other data types within a fast-paced, deadline-oriented environment, in accordance with internal labelling guidelines.

• Evaluating AI-generated content in your native language, assessing linguistic quality, accuracy, and cultural appropriateness for LOTE markets.

• Building and contributing to language-specific datasets to support the internationalisation of Canva AI features.

• Delivering labelled assets based on the set timeline and meeting output targets per labelling task.

• Reviewing and refining your own labels based on feedback to maintain accuracy and consistency.

• Being responsive and consistently checking all communications for task deliveries and updates.

• Managing your own time effectively to ensure timely completion of tasks.

• Being participative and attending all briefing and run-through sessions.

• Following Canva’s data security and confidentiality policies.

• Collaborating with in-house team members for ongoing alignment on quality standards.

Additional Information

What we're looking for

• Experience as a multimedia designer with a strong portfolio showcasing a deep understanding of visual communication fundamentals

• Manage time efficiently and deliver tasks independently

• Exceptional organizational skills and keen attention to detail

• Ability to quickly adapt and learn in a dynamic environment

• Native proficiency in Japanese required

• Excellent written and verbal English communication skills

• Thrive in a fast-paced, deadline-driven setting

• Basic knowledge of data annotation (bonus, but not required)