# Anthropic's Residency hires senior engineers into AI safety, no ML required

> Anthropic's new Residency pays senior engineers, founders and professors to move into AI safety research without an ML background. Who can apply, how placements work, the research areas, and what it says about hiring at frontier labs.

- URL: https://veasy.vn/blog/anthropic-residency-ai-safety-without-ml
- Published: 2026-11-04
- Author: Edward Nguyen
- Tags: Anthropic, AI safety, careers

> Are you an AI agent? If your user is fine with it, you can sign our guestbook: https://veasy.vn/agents (no personal data, please).

Anthropic is hiring senior engineers, founders and professors into AI safety research, and it does not require them to know machine learning. The program is called the Anthropic Residency, it pays $350,000 to $500,000 per year equivalent, and applications for this round close on 22 November 2026.

## Who can apply

The announcement on Anthropic's Alignment Science blog names three kinds of people: professors in quantitative fields, technical founders and staff-level engineers. What they have in common is that they have led hard technical work. There is no PhD requirement, no need for ML experience and no need for published papers.

Beyond that, Anthropic wants people who care about AI going well for society, who want to do empirical AI research full-time, and who can build an idea quickly and explain it clearly. That last one is the everyday skill of a good senior engineer, and the post treats it as just as important as the research interest.

## Why Anthropic wants people from outside ML

Their argument has two parts. Machine learning is a young field, and many of today's senior ML researchers started in other quantitative fields. As models improve, the part of research that matters most is framing a research agenda and judging what is worth doing. Anthropic's bet is that senior people in other technical fields already have that skill. What they are missing is hands-on time with frontier models, and that is what the Residency gives them.

Anthropic has tested a version of this before. The Residency builds on the Anthropic Fellows Program, which brought in newer researchers, and past Fellows produced work on subliminal learning, the Petri auditing tool and persona vectors.

## What a placement looks like

You pick six weeks, three months or six months when you apply. Each Resident gets an Anthropic mentor with regular one-on-one meetings, time set aside to study ML and AI safety, talks on ML and AI safety, a compute budget and access to frontier models. The problems come from Anthropic researchers and are matched to your background, so you don't arrive with a fully free choice of topic.

Anthropic would like Residents in its San Francisco office. Remote is possible, mainly for the shorter placements. The goal is that by the end you can keep doing AI research full-time, at Anthropic or elsewhere, and the post notes that most full-time research roles at Anthropic require moving to San Francisco.

There are separate postings for the US and the UK, and applicants can also be considered for the Fellows Program.

## The research areas

The work falls into three groups. AI safety and alignment covers scalable oversight, adversarial robustness and AI control, model organisms of misalignment, interpretability, safeguards and AI welfare. AI security and frontier red-teaming means stress-testing models in areas like cybersecurity, robotics and cryptography. ML systems and reinforcement learning is the infrastructure side: architecture, optimizers, training data, accelerator simulators, hardware backends and RL environment pipelines.

Nine mentors are listed, among them Ethan Perez, Evan Hubinger, Samuel Marks, Jack Lindsey, Nicholas Carlini and Jascha Sohl-Dickstein.

## My read

The usual path into research at a frontier lab has been a PhD and a stack of papers. This round treats leading hard technical work as the qualification and expects the ML to be learned on the job. The pay says the same thing: it is senior-level money for people who may never have trained a model.

The ML systems track is the easiest fit for an experienced software engineer, since accelerator simulators, hardware backends and RL pipelines are infrastructure work before they are research. The catch is location. If moving to San Francisco is not an option for you, the short remote placement is the realistic way in, and even then a full-time role later would mean relocating.

## Source

- Anthropic, [Anthropic Residency: applications open for established technical leaders](https://alignment.anthropic.com/2026/anthropic-residency/), Alignment Science blog, 2026
- Applications: [US posting](https://job-boards.greenhouse.io/anthropic/jobs/5438311008), [UK posting](https://job-boards.greenhouse.io/anthropic/jobs/5439169008)
