포지션 상세
Keep AI Under Control.
AI now makes decisions and takes action on its own. It executes code, accesses enterprise systems, and moves through the physical world through robots. As we entrust AI with greater capabilities and authority, technology that verifies and controls its behavior becomes essential.
AIM Intelligence works at the frontier of AI safety and security, from models to agents to robots. We uncover where and how the most advanced AI systems fail, and turn those findings into defenses that work in the real world. Our automated red teaming platform Stinger finds vulnerabilities and risks, and our real-time security platform Starfort controls data, permissions, and AI behavior. Our research and engineering span everything from model responses to agents' tool use to robots' physical actions.
We are already solving these problems with world-leading AI developers and in real industrial settings. We have contributed to safety projects with global AI companies such as OpenAI and Anthropic, and presented our research at leading AI conferences. More than 30 major Korean enterprises are our customers, and since our founding in 2024 we have raised over KRW 13 billion in cumulative funding from investors including Samsung Venture Investment, LG Electronics, Hyundai Motor Group's ZER01NE Ventures, NAVER D2SF, and Mirae Asset.
Founded in Korea, AIM aims to lead global AI safety and security with technology that verifies and controls AI. Building on our collaborations with global AI companies, we are expanding our international customers and research partners, and turning new security technologies for agents and Physical AI into products. The problems we take on, the markets we compete in, and the colleagues we work with span the globe.
At AIM, you take direct ownership of important problems that don't have answers yet. You research new attacks and evaluation methods, make products work in complex customer environments, and design experiences that help people understand and control AI. On a team where research, product, design, and business are tightly connected, you turn your own discoveries and judgment into real outcomes.
We're looking for colleagues who dig deep, build hands-on, and verify all the way through. We're waiting for people who learn as fast as AI evolves and are ready to stake their expertise on problems that have never existed before.
Together, we build the technology to control AI, so that AI can do greater things.
We offer industry-leading compensation that reflects your skills, impact, and contributions.
■About the Role
Build the systems that reveal how AI-powered robots fail. We are looking for a Robotics Software Engineer who will reproduce unsafe robot behavior through robot integrations, simulation, and physical testing, and turn it into repeatable safety evaluations for Physical AI.
You will collaborate with our Physical AI researchers and Stinger engineers to turn risk scenarios and evaluation criteria into executable tests. Your work will connect simulation and physical robot environments, make test outcomes traceable and reproducible, and develop prototypes into reusable evaluation tools and product features.
Hands-on experience integrating, operating, and debugging physical robots is central to this role. You do not need experience with every type of robot; we value the ability to understand a robotic system and make its software, communication, and control interfaces work reliably together.
•Work with researchers to implement risk scenarios and evaluation criteria as executable tests, including the conditions needed to reproduce and assess a result.
•Build prototypes and demonstrations that show how robot agents respond to adversarial inputs and how those responses affect their behavior.
•Capture and connect model responses, action plans, control commands, execution logs, and observed outcomes so the team can investigate failures and verify results.
•Debug integration, communication, and execution issues. Automate repeat runs, result comparisons, and regression checks as evaluation capabilities develop.
•Turn experimental code and demonstrations into reusable integration modules, test tools, and technical documentation for Stinger.
•Set up physical test environments and apply safe operating procedures when running experiments on real robots.
•Share failure cases and evaluation findings with research and product teams to help improve safety evaluation and guardrails.
•Programming skills to build robot integrations, automate tests, collect data, and debug system behavior.
•The ability to investigate software, communication, and control issues that arise when connecting and operating robots.
•Experience defining an approach to an open-ended technical problem, implementing it, and verifying the result through testing.
•The ability to document test conditions, results, and limitations clearly, and collaborate with researchers and engineers to deliver working software.
•Eligible for international business travel.
AI now makes decisions and takes action on its own. It executes code, accesses enterprise systems, and moves through the physical world through robots. As we entrust AI with greater capabilities and authority, technology that verifies and controls its behavior becomes essential.
AIM Intelligence works at the frontier of AI safety and security, from models to agents to robots. We uncover where and how the most advanced AI systems fail, and turn those findings into defenses that work in the real world. Our automated red teaming platform Stinger finds vulnerabilities and risks, and our real-time security platform Starfort controls data, permissions, and AI behavior. Our research and engineering span everything from model responses to agents' tool use to robots' physical actions.
We are already solving these problems with world-leading AI developers and in real industrial settings. We have contributed to safety projects with global AI companies such as OpenAI and Anthropic, and presented our research at leading AI conferences. More than 30 major Korean enterprises are our customers, and since our founding in 2024 we have raised over KRW 13 billion in cumulative funding from investors including Samsung Venture Investment, LG Electronics, Hyundai Motor Group's ZER01NE Ventures, NAVER D2SF, and Mirae Asset.
Founded in Korea, AIM aims to lead global AI safety and security with technology that verifies and controls AI. Building on our collaborations with global AI companies, we are expanding our international customers and research partners, and turning new security technologies for agents and Physical AI into products. The problems we take on, the markets we compete in, and the colleagues we work with span the globe.
At AIM, you take direct ownership of important problems that don't have answers yet. You research new attacks and evaluation methods, make products work in complex customer environments, and design experiences that help people understand and control AI. On a team where research, product, design, and business are tightly connected, you turn your own discoveries and judgment into real outcomes.
We're looking for colleagues who dig deep, build hands-on, and verify all the way through. We're waiting for people who learn as fast as AI evolves and are ready to stake their expertise on problems that have never existed before.
Together, we build the technology to control AI, so that AI can do greater things.
We offer industry-leading compensation that reflects your skills, impact, and contributions.
■About the Role
Build the systems that reveal how AI-powered robots fail. We are looking for a Robotics Software Engineer who will reproduce unsafe robot behavior through robot integrations, simulation, and physical testing, and turn it into repeatable safety evaluations for Physical AI.
You will collaborate with our Physical AI researchers and Stinger engineers to turn risk scenarios and evaluation criteria into executable tests. Your work will connect simulation and physical robot environments, make test outcomes traceable and reproducible, and develop prototypes into reusable evaluation tools and product features.
Hands-on experience integrating, operating, and debugging physical robots is central to this role. You do not need experience with every type of robot; we value the ability to understand a robotic system and make its software, communication, and control interfaces work reliably together.
주요업무
•Integrate AI models and agents with robot SDKs, control interfaces, and simulation or physical test environments.•Work with researchers to implement risk scenarios and evaluation criteria as executable tests, including the conditions needed to reproduce and assess a result.
•Build prototypes and demonstrations that show how robot agents respond to adversarial inputs and how those responses affect their behavior.
•Capture and connect model responses, action plans, control commands, execution logs, and observed outcomes so the team can investigate failures and verify results.
•Debug integration, communication, and execution issues. Automate repeat runs, result comparisons, and regression checks as evaluation capabilities develop.
•Turn experimental code and demonstrations into reusable integration modules, test tools, and technical documentation for Stinger.
•Set up physical test environments and apply safe operating procedures when running experiments on real robots.
•Share failure cases and evaluation findings with research and product teams to help improve safety evaluation and guardrails.
자격요건
•Hands-on experience developing, integrating, or operating physical robots or autonomous systems, including direct contributions to the software involved.•Programming skills to build robot integrations, automate tests, collect data, and debug system behavior.
•The ability to investigate software, communication, and control issues that arise when connecting and operating robots.
•Experience defining an approach to an open-ended technical problem, implementing it, and verifying the result through testing.
•The ability to document test conditions, results, and limitations clearly, and collaborate with researchers and engineers to deliver working software.
•Eligible for international business travel.
