Return to jobs
Join the cutting-edge field of AI safety with Apollo Research, focused on mitigating the risks of deceptive alignment in AI
Overview
Salary
£100000 - £200000
Location
London, UK
Expires
Expires at anytime
Apollo Research is a pioneering company in the realm of AI safety. We are uniquely committed to mitigating the risks posed by deceptive model alignment - a rapidly advancing and critically important field. We work hand-in-hand with leading AI organisations globally.
Role Summary:
- Develop and maintain tools for AGI safety research
- Work on projects including the creation of internal tools, development of evaluation pipelines, and orchestrating high-security evaluations
- Optimize CI pipelines, develop telemetry APIs, and enhance the Inspect framework
Role Requirements:
- Requires 5+ years of professional software engineering experience
- Proficiency in Python coding
- Experience in leading software development projects and working with LLM agents or evaluations
- Knowledge in infosecurity or cybersecurity is desirable, as is an interest in AI Safety
Benefits:
- Competitive salary range from 100k to 200k GBP
- Flexible work hours, unlimited vacation and sick leave, free meals on workdays
- Paid work trips, professional development budget, potential visa sponsorship for qualified candidates
- This is a full-time, on-site role in London, with possible partial remote work in exceptional cases
ABOUT APOLLO RESEARCH Apollo Research is focused on the rapid advancements in AI capabilities, which while promising, also pose significant risks. We are particularly concerned with deceptive alignment or scheming, where a model may seem aligned but is actually misaligned and capable of bypassing human oversight. Our work includes building evaluations, studying the science of scheming, and developing mitigations against such risks. We collaborate with leading AI companies to test their models before deployment and work on scheming mitigations.
THE ROLE As a Backend Software Engineer at Apollo Research, you will be tasked with building tools for frontier AGI safety research. This includes maintaining and developing libraries and tools for monitoring and controlling LLM traffic. You will be involved in projects such as creating internal tooling for running and analyzing evaluations, developing automated evaluation pipelines, and orchestrating tools for high-security, reliable parallel evaluations. Your role will also include optimizing CI pipelines, developing telemetry APIs, and improving the Inspect framework.
JOB REQUIREMENTS Candidates should have a minimum of 5 years of professional software engineering experience and must be proficient in writing production-quality Python code. Experience in leading software development projects, working with LLM agents or evaluations, and knowledge in infosecurity or cybersecurity are highly desirable. Interest in AI Safety is also a plus.
BENEFITS Apollo Research offers a competitive salary range of 100k to 200k GBP, flexible work hours, unlimited vacation and sick leave, and free meals on workdays. Additional benefits include paid work trips, a yearly professional development budget, and the possibility of visa sponsorship for qualified candidates.
ADDITIONAL INFORMATION This is a full-time, on-site position located next to the London Initiative for Safe AI offices in London. In exceptional cases, partially remote arrangements may be considered. We are committed to equality and value diversity at our company.
