Research Scientist, Agent Robustness

Scale Labs, Research Scientist — Frontier Risk Evaluations As the leading data and evaluation partner for frontier AI companies, Scale plays an integral role in understanding the capabilities and safeguarding AI models and systems. Building on this expertise, Scale Labs has launched a new team focused on policy research, to bridge the gap between AI research and global policymakers to make informed, scientific decisions about AI risks and capabilities. Our research tackles the hardest problems in agent robustness, AI control protocols, and AI risk evaluations to help governments, industry, and the public understand and mitigate AI risk while maximizing AI adoption. This team collaborates broadly across industry, the public sector, and academia and regularly publishes our findings. We are actively seeking talented researchers to join us in shaping this vision. As a Research Scientist focused on Frontier Risk Evaluations, you will design and create evaluation measures, harnesses and datasets for measuring the risks posed by frontier AI systems. For example, you might do any or all of the following: Design and build harnesses to test AI models and systems (including agents) for dangerous capabilities such as security vulnerability exploitation, CBRN uplift, and other high-risk activities; Work with government agencies or other labs to collectively scope and design evaluations to measure and mitigate risks posed by advanced AI systems; Publish evaluation methodologies and write technical reports for policymakers. Ideally you’d have: Commitment to our mission of promoting safe, secure, and trustworthy AI deployments in the industry as frontier AI capabilities continue to advance. Practical experience conducting technical research collaboratively. You should be comfortable building and instrumenting ML pipelines, writing evaluation harnesses, and quickly turning new ideas from the research literature into working prototypes. A track record of published research in m

Scale Labs, Research Scientist — Safety Post Training As the leading data and evaluation partner for frontier AI companies, Scale plays an integral role in understanding the capabilities and safeguarding AI models and systems. Building on this expertise, Scale Labs has launched a new team focused on policy research, to bridge the gap between AI research and global policymakers to make informed, scientific decisions about AI risks and capabilities. Our research tackles the hardest problems in agent robustness, AI control protocols, and AI risk evaluations to help governments, industry, and the public understand and mitigate AI risk while maximizing AI adoption. This team collaborates broadly across industry, the public sector, and academia and regularly publishes our findings. We are actively seeking talented researchers to join us in shaping this vision. As a Research Scientist working on Safety Post-Training you will develop and apply post-training methods and interpretability techniques to make frontier AI systems safer, and better understood by researchers and policymakers.. For example, you might: Design and run post-training pipelines to study how training choices affect model safety, robustness, and alignment properties; Develop interpretability-informed evaluations that reveal how and why models produce unsafe, deceptive, or otherwise undesirable behaviors, and use those insights to guide targeted mitigations; Collaborate with policymakers, engineers, and other researchers to translate post-training and interpretability findings into actionable safety standards, evaluation benchmarks, and best practices. Ideally you’d have: Commitment to our mission of promoting safe, secure, and trustworthy AI deployments in the industry as frontier AI capabilities continue to advance. Experience with post-training and RL techniques such as RLHF, DPO, GRPO, and similar approaches. A track record of published research in machine learning, particularly in generati

Scale Labs, Research Scientist — AI Controls and Monitoring As the leading data and evaluation partner for frontier AI companies, Scale plays an integral role in understanding the capabilities and safeguarding AI models and systems. Building on this expertise, Scale Labs has launched a new team focused on policy research, to bridge the gap between AI research and global policymakers to make informed, scientific decisions about AI risks and capabilities. Our research tackles the hardest problems in agent robustness, AI control protocols, and AI risk evaluations to help governments, industry, and the public understand and mitigate AI risk while maximizing AI adoption. This team collaborates broadly across industry, the public sector, and academia and regularly publishes our findings. We are actively seeking talented researchers to join us in shaping this vision. As a Research Scientist focused on AI Controls and Monitoring, you will design methods, systems, and experiments to ensure that advanced AI models and agents remain aligned with intended goals, even in high-stakes or adversarial environments. For example, you might: Develop monitoring techniques and observability methods that track AI behavior in real time to identify and flag deviations, emergent capabilities, or anomalous outputs; Research mechanisms for layered control, including fail-safes, oversight protocols, and intervention methods that can halt or redirect AI systems when risks are detected; Design red-team simulations to probe weaknesses in oversight and control mechanisms, and build mitigations to close identified gaps; Collaborate with policymakers, engineers, and other researchers to establish standards and benchmarks for AI monitoring and escalation. Ideally you’d have: Commitment to our mission of promoting safe, secure, and trustworthy AI deployments in the industry as frontier AI capabilities continue to advance. Practical experience conducting technical research collaboratively. You should be

Scale works with the industry's leading AI labs to provide high quality data and accelerate progress in GenAI research. We are looking for Research Scientists and Research Engineers with expertise in LLM post-training (SFT, RLHF, reward modeling) and evaluation. This role is on the evaluation pod within the GenAI Research Organization and will focus on building benchmarks and diagnosing model failure modes in both text and multimodal modalities. In this role, you will develop rigorous evaluations and diagnostic methods that reveal where frontier models fail and why. You will collaborate with researchers and engineers to define best practices in evaluation-driven AI development. You will also partner with top foundation model labs to translate failure analysis into technical and strategic input on the next generation of generative AI models. You will: Analyze model behavior to identify, characterize, and diagnose failure modes in frontier LLMs and Agents. You’ll identify everything from capability gaps and reasoning errors to robustness and alignment issues, all focusing on RCA. Design and build benchmarks and evaluation methods that measure LLM capabilities in both text and multimodal modalities. Apply post-training expertise (SFT, RLHF, reward modeling) to connect observed failures to the data and training interventions that address them. Publish research findings in top-tier AI conferences. Ideally you’d have: Ph.D. or Master's degree in Computer Science, Machine Learning, AI, or a related field. Deep understanding of deep learning, reinforcement learning, and large-scale model fine-tuning. Experience with post-training techniques such as RLHF, preference modeling, or instruction tuning, and with LLM evaluation or benchmark development. Excellent written and verbal communication skills. Published research in areas of machine learning at major conferences (NeurIPS, ICML, ICLR, ACL, EMNLP, CVPR, etc.) and/or journals. Previous experience in a customer facing r

About Scale At Scale AI, our mission is to accelerate the development of AI applications. For 8 years, Scale has been the leading AI data foundry, helping fuel the most exciting advancements in AI, including: generative AI, defense applications, and autonomous vehicles. With our recent Series F round, we’re accelerating the abundance of frontier data to pave the road to Artificial General Intelligence (AGI), and building upon our prior model evaluation work with enterprise customers and governments, to deepen our capabilities and offerings for both public and private evaluations. About the ACE team The Agent Capabilities & Environments (ACE) team, part of Scale’s Research organization, brings together customer-facing Researchers and Applied AI Engineers. Our core mission includes research on agent environments and RL reward signals, benchmarking autonomous agent performance across real-world scenarios and environments, creating robust data programs to improve Large Language Models (LLMs) agentic capabilities and building foundational tools and frameworks for evaluating models as agents. ACE focuses on autonomous agents that dynamically interact with diverse external environments, including code repositories, GUI interfaces, browsers, and more. About This Role This role is at the intersection of cutting-edge AI research and practical application, with a focus on studying the data types essential for building state-of-the-art agents, such as browser and SWE agents. The ideal candidate will explore the data landscape needed to advance intelligent, adaptable AI agents, guiding the data strategy at Scale to drive innovation. This position requires not only expertise in LLM agents and planning algorithms but also creativity in addressing novel challenges related to data, interaction, and evaluation. You will contribute to impactful research publications on agents, collaborate with customer researchers, and work alongside the engineering team to translate t

Scale’s ML platform (RLXF) team builds our internal distributed framework for large language model training and inference. The platform has been powering MLEs, researchers, data scientists and operators for fast and automatic training and evaluation of LLM's, as well as evaluation of data quality. Scale is uniquely positioned at the heart of the field of AI as an indispensable provider of training and evaluation data and end-to-end solutions for the ML lifecycle. You will work closely across Scale’s ML teams and researchers to build the foundation platform that supports all our ML research and development. You will be building and optimizing the platform to enable our next generation of LLM training, inference and data curation. If you are excited about shaping the future AI via fundamental innovations, we would love to hear from you! You will: Build, profile and optimize our training and inference framework Collaborate with ML teams to accelerate their research and development and enable them to develop the next generation of models and data curation Research and integrate state-of-the-art technologies to optimize our ML system Ideally you’d have: Strong excitement about system optimization Experience with multi-node LLM training and inference Experience with developing large-scale distributed ML systems Strong software engineering skills, proficient in frameworks and tools such as CUDA, Pytorch, transformers, flash attention, etc. Strong written and verbal communication skills and the ability to operate in a cross functional team environment Nice to haves: Demonstrated expertise in post-training methods &/or next generation use cases for large language models including instruction tuning, RLHF, tool use, reasoning, agents, and multimodal, etc. Compensation packages at Scale for eligible roles include base salary, equity, and benefits. The range displayed on each job posting reflects the minimum and maximum target for new hire salaries for the positi