Member of Technical Staff, Research

Meridian Labs builds open source tools for evaluating and understanding AI models, including Inspect AI, Inspect Scout, Inspect Petri, and Inspect Flow. These tools are used by AI Security Institutes (AISIs) in the UK, US, EU, Japan, and Korea, and by research organizations such as METR, Apollo Research, Epoch AI, and Redwood Research. We also use them ourselves: Petri alignment audits run by our team have appeared in Anthropic model cards.

As AI systems become more capable and autonomous, the tools used to evaluate them and investigate what they do have to keep up. Investigating a real AI incident, for example, can now mean making sense of thousands of agent transcripts, each spanning days of activity, far more than anyone can read. Researchers have to lean on AI tools for the analysis, but those tools often miss key details, get things wrong, or overstate their confidence. We’re looking for someone with strong research and engineering skills to build tools that researchers can trust across evaluations, transcript analysis, and incident investigations.

Our team works closely with researchers at government AI Security Institutes and other organizations that use our tools, and many of us work directly with them. You’ll see your work used in real evaluations and audits, and hear quickly when something falls short.

Responsibilities

  • Work on projects like these:
    • Improve tools and methods for automated alignment evaluation, from Petri’s auditor strategies to detecting concerning behaviors
    • Build and research open source monitors to watch for agent misbehavior in evals and day-to-day usage
    • Build tools that make transcript analysis and incident investigations faster, and their findings more trustworthy
  • Design and run experiments to test tooling and methods
  • Ship and maintain tools that come out of this work in the Inspect, Petri, and Scout ecosystem
  • Collaborate directly with researchers at AISIs and partner organizations on their evaluations and investigations

Minimum Qualifications

  • Strong Python skills and comfort working in existing codebases
  • Hands-on experience working with LLMs: prompting, evaluation, or similar
  • Experience in AI safety or security research: capability or alignment evaluations, red teaming, pre-deployment testing, or similar
  • Experience designing and running experiments: you pick sensible baselines and controls, sanity-check your results, look for confounds, and report uncertainty honestly
  • Attention to detail with model outputs: you read the transcripts, not just the summary statistics, and you notice when something looks off
  • Clear writing for both technical and non-technical readers, and comfort working directly with researchers outside our team

Preferred Qualifications

  • Published research or other public work in AI safety, security, or alignment, such as papers, blog posts, or open source projects
  • Experience measuring LLM judge reliability, such as agreement with human labels, calibration, or bias
  • Experience with Inspect AI, Petri, Scout, or similar tools
  • Experience analyzing large volumes of agent transcripts or logs, or investigating incidents in complex systems

If you don’t meet every qualification, we’d still like to hear from you.

Location

Meridian Labs is based in Boston, MA. We prefer in-person or hybrid work in Boston, the San Francisco Bay Area, or London, and we’re open to remote for the right person.

Compensation

$175,000 – $325,000 with benefits, depending on experience and level.

Applying

To apply, please email the following to jobs@meridianlabs.ai:

  • Your resume
  • LinkedIn URL (if available)
  • GitHub URL (if available)
  • Links to research or evaluation work you’ve done (if available)
  • A short note on why you’re interested in joining Meridian Labs
  • How you learned about this role