Tianyi Alex Qiu

I build solutions to AI alignment, with a focus on how it uplifts human truth-seeking and moral progress - what I believe to be the most important problem of our time. Research projects I led on this front have received two Best Paper Awards at leading machine learning venues. Read more about me.

• The drive and willingness to learn everything on-the-fly; best if you are historically good at self-teaching.
• If you pick/propose a technical project, you must have at least some amount of technical experience or background, whether formal or informal.
• If you pick/propose a technical project with heavy ML experimentation elements, you must be at least familiar with Python and Linux. Exceptions are possible if you have a compelling case for your fit, so please don't hesitate to apply if that's the case.

By default, the mentee is expected to take ownership of the idea they choose while we give guidance/feedback (and as a result, they will be the first author of any publication resulting from such a mode of collaboration), unless someone is already working on it, in which case we will see what type of collaboration mode makes the most sense. This is less daunting than it may sound; even people who are completely new to research can do reasonably well in leading a project, as long as they have the relevant skillset & pick a tractable project.

This depends heavily on the exact project chosen, but common tasks include:

1. running ML / LLM experiments
2. designing ML / LLM experiments with our guidance
3. result analysis and visualization
4. running human subject experiments
5. building public-facing web demos
6. writing papers

• Guiding direction
• Technical guidance (at ~2hr/wk capacity)
• Facilitating connections
• Feedback on anything