Test an AI safety idea in a weekend

My life changed when I decided to apply for a deception detection hackathon run by Apart Research. I am genuinely grateful for the work Apart has done, and have donated a portion of my own salary to Apart so that more researchers like me can be supported and emboldened to make the leap to working on humanity’s greatest problems.

Today, because of Apart Research’s initial spark, I’m thriving in a full-time AI safety research role, having left my engineering position behind after graduation. From being a curious engineering student to a published researcher with a dedicated mentor, my career trajectory has been utterly transformed in just a few months.

Thanks to the Apart hackathon on evals with METR I got a job as contractor at METR. Being part of the Apart community opened doors to people with a range of relevant experience… Our project’s acceptance at a NeurIPS workshop added credibility and gave me helpful exposure to academic writing and how ML projects are structured.

During an Apart Research hackathon I got my first hands-on experience with mechanistic interpretability and fell in love with it. Now I’m working on mechinterp full-time and have published mechinterp papers at ICLR and ICML. Generally their hackathons are very accessible, well-organized, and a great entry-point for people interested in AI safety.

Helped me learn about mech interp for the first time, find a collaborator and write my first research blog post. I now work full time as a mech interp researcher, and Apart hackathons substantially accelerated / partly caused this happening.

I wouldn’t have gotten into AI Safety if not for Apart’s sprints… I would never have considered submitting a workshop paper to an A* conference, let alone being an equal first author on an ICLR oral. Apart’s trust in its hackathon participants, based solely on merit, regardless of credentials, was significant. It was the launchpad I needed.

I first joined Apart through a hackathon, and from there, I met a lot of my collaborators and a community of talented and ambitious researchers. Apart helped me build foundational research skills, gain confidence, and engage deeply with the broader AI safety community.

I participated in an Evals hackathon for over a weekend and I loved it. I made a great friend and we published a paper together. That experience helped me build my portfolio and develop relevant LLM skills.

An Apart hackathon was the first practical thing I did in AI safety. Much of AI safety introduction involves mostly reading, but doing practical work is arguably more important. An Apart hackathon is the best possible introduction to this.

Participating in Apart’s hackathon was a pivotal first step toward my career in AI Safety research. This publication significantly strengthened my portfolio, directly helping me secure interviews for research internships in AI Safety. At least $3,000 is 100% attributable to the support from Apart.