Research Engineer, Model Evaluation and Improvement

Benchling • San Francisco, CA

4d • $136,435 - $166,754 • Onsite

About The Position

Benchling is seeking a Research Engineer to help make frontier AI models better at science. The role focuses on building datasets, evaluations, and systems that address the gap between models’ existing biological knowledge and their ability to reason through real-world scientific problems. The engineer will work with scientists to translate complex scientific work and expert judgment into rigorous tasks and evaluation criteria, and collaborate with leading AI labs to understand model failures and improve performance.

This position sits at the intersection of software engineering, biology, and frontier AI. It involves identifying tasks that are challenging for LLMs and valuable to scientists, designing evaluations that capture scientific judgment, and building systems to create tasks at scale. Benchling prioritizes in-office collaboration; employees are expected to work on-site five days per week, Monday through Friday.

Requirements

• At least 2 years of academic or industry experience at the intersection of biology and AI, evaluating and improving scientific models or LLMs for biological applications.
• Experience building with LLMs and an understanding of where current models excel and struggle, and how to design systems around their capabilities.
• Curiosity about frontier AI and a desire to understand and advance the capabilities of rapidly improving models.
• Comfort working on ambiguous problems in a rapidly changing area where technical approaches are still emerging.
• Collaborative mindset and ability to work closely with engineers, scientists, and external research partners.
• Desire to work in a fast-paced environment where priorities can shift and rapid experimentation is encouraged.

Responsibilities

• Build datasets to evaluate and improve frontier models by converting complex scientific data into high-quality tasks and environments for LLMs.
• Analyze model failure modes by running experiments across frontier models and identifying where they struggle and opportunities for improvement.
• Build scalable data infrastructure and pipelines to curate, transform, and validate scientific data for model evaluation and improvement.
• Collaborate with frontier AI labs to develop and evaluate approaches for improving models on challenging scientific tasks.
• Work with scientists to translate expert judgment into problems and evaluation criteria that distinguish strong model behavior.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed