Staff Research Engineer, Frontier AI Evaluation

OpenTrain AI, Inc.; United States; Remote; USD 103,000 - 193,000; Part time

Job summary

OpenTrain AI, Inc. seeks a Staff Research Engineer (part-time contractor, ~20 hours/week) to investigate capabilities, limits, and training methods of advanced AI systems.

You’ll blend research with engineering to translate findings into practical AI applications, working across datasets, RL environments, evaluation, synthetic data, and model understanding. You will collaborate with research, engineering, product, and operations teams, publish findings, mentor, and participate in peer

Qualifications

• PhD or master's in artificial intelligence, machine learning, computer science, or closely related field, or equivalent research experience.
• At least 7 years of professional experience including substantial research engineering work in ML or frontier AI systems.
• Strong Python programming and experience with modern ML frameworks.

Responsibilities

• Formulate research questions and design experiments to produce evidence-based conclusions.
• Build datasets, prototypes, tools, benchmarks, and evaluation frameworks for complex multi-step workflows.
• Train, test, and evaluate models with ML tools, then analyze results.

Skills

Python

Experiment design

Communication

Independent work

Mentoring

Education

PhD in AI/ML/CS

Master's in AI/ML/CS

Tools

TensorFlow

PyTorch

JAX

Job description

The Work

As a Staff Research Engineer, you will investigate the capabilities, limits, and training methods of advanced AI systems. You will combine research and engineering to turn reliable findings into practical AI applications.

You will work across research-grade datasets, reinforcement learning environments, model evaluation, synthetic and agentic data generation, benchmarks, and model understanding. The work may include coding-agent, computer-use, browser-use, and function-calling systems.

• Formulate research questions and design experiments that produce evidence-based conclusions.
• Build datasets, prototypes, tools, benchmarks, and evaluation frameworks for complex, multi-step workflows.
• Train, test, and evaluate models with modern machine learning tools, then analyze the results.
• Explore reinforcement learning, post-training, synthetic data, agentic systems, model understanding, and AI evaluation.
• Collaborate with research, engineering, product, and operations stakeholders.
• Share findings through technical reports or other appropriate channels, and contribute to peer review, technical discussions, mentoring, and research collaboration.

What It Pays And Takes

The listing does not state a pay rate. This is a part-time contractor role for an expert-level candidate, with a commitment of at least 20 hours per week.

• Pay: Not provided in the listing.
• Workload: 20+ hours per week.
• Location: Worldwide.
• Language: English.
• Education: A PhD or master's degree in artificial intelligence, machine learning, computer science, or a closely related technical field, or equivalent research experience.
• Experience: At least 7 years of professional experience, including substantial research engineering work in machine learning or frontier AI systems.
• Technical skills: Strong Python programming and experience with modern artificial intelligence and machine learning frameworks.
• Research skills: Experience designing experiments, training or evaluating models, or developing AI systems.
• Working style: Sound judgment about experimental rigor and reproducibility, clear communication, independent execution, cross-functional collaboration, and technical mentoring experience.
• Valuable background: Synthetic or agentic data generation, reinforcement learning, post-training, model understanding, AI evaluation, benchmarks, agents, tool-using systems, coding-agent environments, user-interface or browser-use environments, function-calling systems, technical publications, open-

About AI Training Work

AI training is the human work behind modern AI systems, including preparing datasets, testing models, rating outputs, and building evaluation methods. Experienced researchers are paid to design reliable experiments, improve data and evaluation quality, and help determine how advanced systems perform in real tasks.