SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
ServiceNow AI Research is seeking a Staff Applied Research Scientist to work on foundational problems in agentic AI systems operating at enterprise scale. In this role, you will tackle vaguely defined challenges where no established playbook exists, transforming ambiguous problems into concrete, testable research directions.
Your core responsibilities include:
- Identifying high-impact research questions and proposing models and techniques to address them
- Building, training, and evaluating AI agent systems to validate whether proposed approaches work in practice
- De-risking new concepts through rapid prototyping and hypothesis stress-testing before broader organizational commitment
- Collaborating closely with the Forward Deployed Engineering team to validate research ideas against real customer workflows and production environments
- Diagnosing why prototypes fail in the field and determining next steps based on empirical findings
- Shaping the research agenda, product roadmap, and evaluation practices through your findings
- Setting technical direction for applied research projects within the team's broader program
- Mentoring researchers and interns
- Representing the work through publications, conference talks, and open-source contributions
You will work at the intersection of research rigor and practical impact, translating findings into clear recommendations for product and executive audiences.
REQUIREMENTS:
- PhD in computer science, machine learning, or related field, or equivalent research/applied experience
- Three or more years of post-PhD experience, including work on production systems or systems close to real users
- Ability to decompose ill-defined problems into components, form hypotheses, and design experiments to test them
- Hands-on experience implementing, training, and evaluating LLM-based systems and agent architectures, including tool use, planning, and multi-step workflows
- Experience designing evaluations: defining metrics, building test sets, and analyzing failure modes
- Working knowledge of AI security and safety issues (prompt injection, data leakage, misuse)
- Strong software development skills in Python and modern ML frameworks, with reproducible and maintainable code
- Comfort working directly with engineers and customer-facing teams; ability to translate findings into clear recommendations
- Experience mentoring researchers, engineers, or interns
- Advanced use of agentic tools in your own workflow
NICE TO HAVE:
- Publications or open-source contributions in agent evaluation, AI security, or agent reliability
- Experience with AI systems in enterprise environments (compliance, governance, scale constraints)
- Experience building benchmarks, red-teaming pipelines, or monitoring deployed models
- Background in systems engineering, security research, or formal methods
