The Anthropic AI residency is offering established technical leaders a route into AI safety research without requiring prior machine learning experience. The paid program runs from six weeks to six months , with annualized compensation of $350,000 to $500,000 . Applications close on November 22, 2026.
Key takeaways
Professors, technical founders and staff-level engineers are among the intended applicants.
Residents receive individual mentorship, frontier-model access and a dedicated compute budget.
Remote participation is possible, particularly for shorter placements.
The program aims to prepare participants for full-time AI research at Anthropic or elsewhere.
Who the Anthropic AI residency is designed for
Anthropic is seeking people who have already led ambitious technical work and want to move into empirical AI research. Its target applicants include professors in quantitative disciplines, technical founders and staff-level engineers.
No PhD, machine learning background or published papers are required. Instead, Anthropic emphasizes technical leadership, motivation to make AI safe and beneficial, clear communication and the ability to implement ideas quickly in collaborative environments.
Placements come in three lengths: six weeks, three months or six months. Applicants can indicate their preferred duration, and Anthropic says it will aim to accommodate flexible start dates. Pay reflects seniority and is prorated to the length of the placement; the quoted range is an annual equivalent, not the total payment for a residency.
Candidates apply through separate US and UK postings . Offers are expected in January 2027. Anthropic encourages applications from underrepresented groups in technology and says candidates may also receive consideration for its Fellows Program.
Research, mentorship and working arrangements
Residents will tackle empirical research questions with Anthropic teams, working on problems matched to their interests and experience. Each participant receives regular one-on-one mentorship , alongside dedicated self-study time, talks and guidance on modern machine learning and AI safety.
The research spans scalable oversight, adversarial robustness and AI control, model interpretability, safeguards, AI welfare and model organisms of misalignment. Oversight work seeks ways to keep increasingly capable models helpful and honest; interpretability research examines their internal workings to support more targeted safety interventions.
Security and frontier red-teaming projects will stress-test model capabilities in areas including cybersecurity, robotics and cryptography. ML systems and reinforcement learning work covers training methods and infrastructure, including model architectures, optimizers, training data, accelerator simulators, hardware backends and reinforcement-learning environment pipelines.
Residents get access to frontier models and a dedicated compute budget . Anthropic prefers participants to work from its San Francisco offices, but allows remote participation, especially for shorter placements. Most of its full-time research roles require relocation to San Francisco.
Potential mentors include Sara Price , who leads Alignment Training; Ethan Perez, who leads Alignment Science; Evan Hubinger, who leads Alignment Stress-Testing; and Samuel Marks, who leads Scalable Oversight.
Other potential mentors are Elana Simon, whose research covers language and biological model interpretability; Belinda Li, who studies language model representations; Jack Lindsey, who leads Model Psychology; Nicholas Carlini, who researches adversarial machine learning and AI security; and Jascha Sohl-Dickstein, who works at the intersection of safety and pretraining.
Building on the Fellows Program
The Anthropic AI residency adapts the Anthropic Fellows Program for established technical leaders. According to Anthropic, the previous program showed how quickly new researchers in the field can make meaningful contributions, pointing to projects such as subliminal learning, the Petri auditing tool and persona vectors as evidence.
The company’s rationale is that senior people in quantitative fields often already know how to frame research agendas and judge which questions deserve attention. The residency is intended to help them develop the frontier-AI understanding needed to apply those skills.
Its stated goal is for every resident to finish in a position to pursue AI research full-time, either at Anthropic or elsewhere. Applications for this round close on November 22, 2026 .
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
