Research Engineer
Help Iliad's alignment researchers and fellows implement and scale ML experiments, turning exploratory code into reliable, efficient larger runs that test what they set out to test.
About Iliad
Iliad is a nonprofit dedicated to advancing foundational AI alignment research. Our research division is made up of several teams working on the mathematical and conceptual foundations of the field, including Learning Theory, Abstractions & Representations, Agency, and Scalable Oversight & Heuristic Arguments. Alongside our own research, we train new alignment researchers through the Iliad Intensive and the Iliad Fellowship, incubate research organizations and an academic journal, and run a range of conferences that bring together researchers working on the hardest problems in AI safety. We're a small and quickly growing team.
About the Role
Help Iliad’s researchers and fellows implement and scale ML experiments. You’ll work closely with them to understand what they want to test, turn small working experiments into reliable, efficient larger runs, and check that the implementation captures the intended experiment. Alongside research support, you’ll maintain useful tools and pursue automation ideas that help researchers make progress.
What you’ll do
- Implement experiments and take exploratory code to larger runs, checking that changes preserve the intended computation and produce results researchers can interpret.
- Debug failures, investigate unexpected behaviour, and identify performance bottlenecks. Measure the effects of changes to understand whether they improve efficiency and reliability.
- Use LLMs for implementation, debugging, and other research tasks, supplying relevant context and checking generated code and explanations against the work at hand.
- Maintain research tools and automation as projects evolve. Between research-support assignments, identify useful improvements and follow through on projects of your own.
What We're Looking For
- Practical ML engineering ability, including implementing experiments, understanding unfamiliar research code, and making it run reliably and efficiently at larger scales.
- Careful debugging and validation habits: investigate causes, check assumptions, and establish that an implementation does what the researcher intends.
- Good judgment about using LLMs, including recognising errors and deciding how to verify their outputs.
- Close, thoughtful collaboration with researchers, with the curiosity to learn a project’s concepts and explain engineering decisions clearly.
- Initiative and useful ideas, alongside willingness to maintain tools and help others work through difficult implementation problems.
Our benefits include:
- Time Off: 25 days paid vacation annually, plus local holidays.
- Flexible Working Hours: We're a distributed organization with employees in a variety of locations and offer flexible working hours.
- Team Retreats: We have occasional in-person team retreats in California and England.