Research Engineer, QC Automation

About this role

An AI infrastructure company seeks a Research Engineer to build end-to-end quality control automation for training data in reinforcement learning pipelines. You'll design metrics, validate datasets, and partner with data vendors to ensure high-quality post-training data while working autonomously in a lean, research-focused engineering team.

What you'll do

• Automate quality control systems for AI training data generated on the platform, emphasizing principled validation over LLM reliance
• Define quality standards, design experiments, and create metrics to evaluate agent outputs and data trajectories
• Build scalable data validation pipelines and sampling strategies for continuous auditing and anomaly detection
• Debug quality issues with data vendors, diagnose agent failure modes, and improve data generation processes
• Translate QC insights into rule-based or model-assisted validation systems integrated into infrastructure tooling
• Reduce data inconsistencies and edge cases by continuously refining standards and vendor feedback loops

What they're looking for

• Python, Docker, and Linux environments
• Data validation pipeline design and end-to-end QA/QC systems
• Benchmark and evaluation design (rubrics, realistic tasks, RL training trajectories)
• Statistics, experiment design, and metric definition
• Reinforcement learning and post-training data pipelines
• Reward signal analysis and failure mode detection
• Cross-functional communication and independent problem-solving in ambiguous environments
• Data quality assessment and measurement frameworks

Benefits

• Visa sponsorship available for eligible candidates
• On-site opportunities in San Francisco or Singapore, or fully remote contractor arrangement
• Work with accomplished engineering team (Olympiad medalists, AI founders, published researchers)
• High autonomy and ownership of critical infrastructure challenges
• Exposure to post-training and reinforcement learning data quality at scale