Concordia AI is committed to identifying and addressing the most pressing safety risks arising from frontier AI development. As large model capabilities advance rapidly and agent frameworks evolve, risks such as AI misuse and loss of control have become urgent scientific challenges. The AI Safety Researcher will join the team to build safety benchmarks, advance red-teaming attack and defense algorithms, and develop quantitative risk modeling, producing top-tier research outcomes.
Responsibilities:
- AI Safety Benchmark Development: design and build capability and safety benchmarks for large models, AI agents and multi-agent systems; curate evaluation datasets, develop automated testing pipelines, conduct systematic evaluations, and produce technical reports or academic papers.
- AI Red-teaming Attack and Defense Algorithm Research: track state-of-the-art adversarial techniques, conduct red-teaming research on large models and agents, discover critical vulnerabilities and risks, and author technical reports or publications.
- AI Risk Modeling: build quantitative risk models for cybersecurity, biological, chemical, harmful manipulation and loss-of-control risks, using evaluation data, expert forecasts and real-world incident case studies.
- AI Risk Mitigation Research: investigate methods in AI alignment, monitoring, interpretability and agent/multi-agent safety; prototype and validate safety solutions.
- Research Translation & Impact: integrate research outcomes into Concordia AI's Frontier AI Risk Monitoring Platform and commercial safety evaluation services, and support safety governance reports, standards recommendations and policy briefs.
Candidates may specialize in one or two of these tracks.
Required qualifications: alignment with Concordia AI's mission and a strong sense of responsibility; bachelor's degree or above; exceptional scientific reasoning skills; strong foundation in data structures and algorithms and proficiency in Python; solid understanding of large model and AI agent architectures, with hands-on experience in model/agent safety evaluation and risk analysis; excellent technical writing skills; strong Chinese reading, writing and spoken skills.
Preferred: degree in Computer Science, AI, Machine Learning or a related field; 2+ years of research or industry experience in large models, AI agents or AI safety; top-tier conference publications in these areas.
Location: Beijing or Singapore.
Application Deadline: 8 November 2026, 23:59 GMT+8
HOW TO APPLY:
Submit the written application form at https://v.wjx.cn/vm/t7Gskw0.aspx. The hiring process has 4 phases: written application form, online written test (including an additional coding test), 2 online interviews, and a 2-day in-person work trial. Detailed information: https://mp.weixin.qq.com/s/QIunhlvvq3OKpIDZE-1ZiA