I’m mapping out how people are entering mechanistic interpretability and to me it seems like there isn’t a single agreed upon route. Some people begin with reproducing classic experiments, some come through theory or causal ML, others build tools, or jump in through bio/RL/vision.

I would appreciate stories that tell:

How did you start?

What worked?

What didn’t?

What would you do differently if starting today?

Are there alternative routes you’ve seen that people underestimate?

All perspectives are welcomed- even partial experiences