Skip to content
Prototype · illustrative data

Welcome back, Tomás.

You're marked open to roles in evaluations and agent safety. Three organizations matching your interests are hiring this month, and two fellowship deadlines land in the next six weeks.

4
Active applications
1
Interviews in progress
10
Open listings
8
People in network

Recommended for you

Matched on your interests, availability and recent projects.

See all
  • Technical AI Safety Fellowship (Cohort 9)

    Northlight Fellowship · Remote + 2-week London residential · Remote

    Fellowship

    Twelve weeks of mentored empirical safety research, ending with a written output and a demo day. Structurally inspired by well-known programmes in the field; independent and unaffiliated with them.

    £4,500 stipend + travelDeadline 3 Nov 2026
  • Agent Safety Engineer

    Harbourline Labs · Remote (Americas) · Remote

    Job

    Build monitoring and interruption layers for tool-using agent deployments.

    CAD 140k–180kDeadline Rolling
  • Alignment Foundations Course (8 weeks, part-time)

    Northlight Fellowship · Online · Remote

    Course

    Facilitated discussion course covering threat models, evaluations, interpretability and governance.

    Free, funded placesDeadline 20 Sep 2026

Pipeline

Where your applications stand.

Saved1
Applied1
Interview1
Offer1
Closed1
Manage in tracker →

People to meet

Aligned interests, open to talk.

  • Dr. Iris Vandermeer

    Research lead, Meridian Alignment Lab — hiring 3 roles

    Hiring

  • Priya Raghunathan

    Programme director — technical AI safety fellowships

    Hiring

  • Daniel Hersh

    Grants lead — technical safety and security funding

    Open to collaborate

Organizations hiring now

Labs, programmes and funders with live listings.

Directory
  • Small empirical lab studying whether oversight methods hold up when the model is smarter than the overseer.

    3 open listings

  • Northlight Fellowship

    Training program

    Intensive technical AI safety programme, inspired in structure by course-and-bootcamp providers such as BlueDot Impact and ML4Good. Independent and unaffiliated.

    2 open listings

  • Policy shop translating frontier-model evaluation results into workable regulatory instruments.

    2 open listings

  • Harbourline Labs

    Research lab

    Applied control research: assume the model is untrusted, then design protocols that still work.

    2 open listings