Note: The job is a remote job and is open to candidates in USA. Cobalt is seeking PhD-qualified machine learning researchers with experience conducting and evaluating original ML research. The role involves producing and reviewing technical reasoning, evaluating model-generated content, designing problems and scoring rubrics, assessing evidence, and contributing expertise to benchmarks, evaluations, and datasets.
Responsibilities
• Produce written reasoning traces on hard ML problems, capturing how you reach a solution rather than only the solution itself
• Draft, review, or revise expert reference answers to technical problems
• Evaluate the accuracy and quality of model-generated technical content
• Compare and rank multiple model responses, and articulate precisely what makes the stronger response stronger
• Author novel problems in your subfield that have verifiable or defensible correct answers
• Synthesize information from papers, derivations, code, experimental logs, and results tables
• Assess whether stated conclusions are supported by the underlying derivation or experimental evidence
• Identify the specific step at which a model's chain of reasoning breaks down, and classify the failure mode
• Design rubrics and partial-credit criteria for scoring multistep technical tasks
• Apply project-specific annotation guidelines, formatting conventions, and quality standards
• Incorporate feedback from reviewers, project leads, and lab research teams
• Contribute subject-matter expertise to benchmark, evaluation, and dataset development projects
Skills
• PhD, completed or in progress, in machine learning, computer science, statistics, mathematics, physics, or a closely related quantitative discipline, with research that is substantially ML focused
• Direct experience authoring, co-authoring, or substantively contributing to at least one ML research output, such as a peer-reviewed paper, preprint, thesis chapter, or comparable technical artifact
• Demonstrated depth in at least one area, for example optimization, reinforcement learning, language model training and post-training, learning theory, probabilistic methods, computer vision, natural language processing, or systems for ML
• Ability to interpret papers, derivations, code, and experimental results
• Excellent technical writing, editing, and information-synthesis skills
• Strong attention to detail and a commitment to factual accuracy
• Ability to work independently and deliver high-quality work within agreed timelines
Benefits
• Expand your influence through evaluation projects, advisory roles, and research collaborations, while deepening your understanding of how frontier models are trained and assessed.
• Collaborate with researchers from leading institutions and labs on high-impact, flexible work.
• Flexible 10 to 40 hour weeks that fit around your research position and your life.
Company Overview
• Cobalt is the data research lab shaping the frontier of artificial intelligence. It was founded in 2026, and is headquartered in San Francisco, California, USA, with a workforce of 2-10 employees. Its website is https://www.gocobalt.ai/.