Research Engineer

About The Role

We are an expanding AI infrastructure startup focused on building core technology for training and evaluating cutting-edge AI agents. Our team includes accomplished individuals such as International Olympiad medalists, serial AI startup founders, and researchers published at prestigious conferences like ICLR and NeurIPS. We seek Research Engineers to help automate agent quality control, develop benchmarks, and generate synthetic data, thereby influencing how AI agents learn and evolve.

Key Responsibilities

• Develop systems to craft new environments, enhance data quality, and convert real-world workflows into defined tasks and benchmarks.
• Create systems to establish, execute, evaluate, and refine agent training environments.
• Design and implement experiments exploring model behavior, failure cases, and data quality concerns.
• Build tools that assist researchers, engineers, and external data providers in producing higher-quality tasks, trajectories, and feedback loops.
• Manage the complete lifecycle of agent training data: from task conception and environment setup to trajectory gathering, evaluation, and validation.
• Collaborate with external vendors to detect bottlenecks and boost the efficiency and quality of the data production pipeline.
• Develop metrics and conduct analyses to verify the effectiveness of tasks, environments, and evaluations for training advanced AI agents.

Qualifications

• 2 to 4 years of relevant engineering experience.
• Proficient with Python programming, Docker containerization, and Linux operating systems.
• Experienced in benchmarks and evaluations, with an understanding of task realism, rubric dependability, environment usability, and trajectory quality for reinforcement learning training.
• Exceptional attention to detail to identify subtle discrepancies in data, model behaviors, or task design.
• Proven ability to independently build tools, pipelines, or research infrastructure with minimal supervision.
• Comfortable working in early-stage startup environments independently under ambiguous and fast-paced conditions.
• Skilled in designing metrics and validation processes.
• Strong quantitative and technical background, demonstrated through competitive programming, research, or personal projects.
• Ability to excel in loosely structured problem contexts and communicate efficiently across different time zones.

Preferred Skills

• Experience or background in reinforcement learning or AI alignment research.
• Familiarity with large-scale data pipelines and working with vendor ecosystems.
• Contributions to open-source ML/AI projects or publications.

Compensation and Benefits

• Annual salary range between $150,000 and $250,000 USD, based on experience.
• Visa sponsorship is offered.
• Opportunity to obtain equity in a well-funded early-stage AI startup.

Location and Working Conditions

This position requires working on-site at our San Francisco, CA office. Candidates must be able to collaborate in person with the team.