About this role
About the Role
Join an AI evaluation and reinforcement learning team building synthetic data pipelines for AI agents. You will turn real-world professional workflows into scalable training tasks and develop ways to measure whether those tasks improve model capabilities.
What You'll Do
- Build pipelines that transform domain-specific workflows into realistic, structured, and challenging synthetic training tasks.
- Collaborate with subject-matter experts to create tasks across professional and technical domains.
- Develop methods and tools to generate, mutate, validate, and improve synthetic tasks.
- Analyze agent performance to understand what tasks teach and where they fail.
- Create metrics for task diversity, realism, learnability, and overall quality.
What We're Looking For
- Two or more years in software engineering, machine learning engineering, or AI research, with relevant experience building data pipelines or ML infrastructure.
- Hands-on experience applying synthetic data research methods to build end-to-end generation pipelines.
- Proficiency in Python and experience working in Linux environments with Docker.
- Understanding of synthetic data quality, evaluation metrics, and the limitations of generated data.
- Experience building evaluation frameworks, benchmarks, or testing environments for AI agents or language models.
- Ability to independently deliver technical projects, build automated data systems, and spot subtle quality issues and edge cases.
- Familiarity with reinforcement learning, agent workflows, or model post-training is useful.
Compensation & Benefits
Salary range: USD 100,000 to 170,000 annually. Visa sponsorship is available.
Location
On-site in Singapore, Singapore.