Senior Research Scientist, AI Agent Evaluation, ServiceNow, Hyderabad

Application ends: November 19, 2026
Apply Now

Job Description

This AI research scientist job in Hyderabad is in ServiceNow’s AI Evaluations and Observability team. You will design methods and benchmarks that test AI agents before they reach customers. You will take each problem from a research question to a shipped feature.

At a glance

  • Position: Senior Research Scientist, Agent Evaluation (regular, reference JB0075773)
  • Where: Knowledge City, Hyderabad, Telangana, India (hybrid)
  • Experience: 5+ years in ML, applied AI, prompt engineering or agentic AI
  • Focus: Benchmarks for agent reasoning, planning, tool use, reliability and safety
  • Posted: 29 September 2026
  • Apply: open until filled (re-checked 19 November 2026)

What the agent evaluation research scientist does

  • Design evaluation methods and benchmarks for agent reasoning, planning, tool use, reliability and safety
  • Use LLM-as-a-judge, trajectory-based and human evaluation
  • Own problems end to end, from research question to prototype to shipped feature
  • Build the pipelines that run agent traces through model-based judges at scale
  • Build the scoring logic that turns raw output into results teams can act on
  • Curate synthetic and real datasets, and check the evaluator itself against human labels

Why this role matters

AI agents now take actions inside business systems. A mistake can cost a customer money or data. Good evaluation catches those failures before release. Measuring the judge itself is still an open research problem, so this role mixes real research questions with production work.

What ServiceNow is looking for

  • 5+ years in ML, applied AI, prompt engineering or agentic AI, including shipping something real to users
  • Strong Python
  • Practical depth in agentic AI and context engineering: planning, reasoning, memory, tool use, retrieval and long context
  • Experience designing evaluation methods, beyond running existing evaluations
  • Hands-on production work with LLM APIs: prompting, structured output, cost and latency trade-offs
  • Clear communication with technical and non-technical people

Nice to have

  • Experience with AI-assisted coding tools

Pay and location

The posting does not state pay. The role is hybrid, based at ServiceNow’s Knowledge City office in Hyderabad.

How to apply

Apply through the official ServiceNow job posting. ServiceNow’s posting does not give a closing date, so the role is open until filled; ResearchJobs.in will re-check this listing on 19 November 2026. Check the posting for location and work-authorisation details before you apply.

See all our artificial intelligence jobs, or browse more research jobs on ResearchJobs.in.

Hiring institution: ServiceNow

Official advertisement: jobs.smartrecruiters.com

How to prepare for this application

  • Prepare one evaluation you designed: what it measured, how you validated it, and what it changed
  • Revise LLM-as-a-judge methods and their known biases, such as position and length bias
  • Be ready to explain how you would measure agreement between an automated judge and human raters
  • Think about how to score each step of an agent trajectory, as well as the final answer
  • Read recent agent benchmarks for tool use and planning, and be ready to discuss their weak points

About ServiceNow

ServiceNow is a US software company based in Santa Clara, California. Its cloud platform runs workflows for IT, HR, customer service and other business teams. It has a large engineering centre in Hyderabad.

We send one confirmation email first. Every alert has an unsubscribe link.