A 26-year-old Google DeepMind researcher explains how he broke his perfectionist streak and got into leadership

Summary

Neel Nanda, 26, who leads the mechanistic interpretability team at Google DeepMind, says he overcame longstanding perfectionism by deliberately saying yes to more opportunities. He forced himself to produce more work (notably a month of daily blog posts) to build visibility, accept risk and increase chances of serendipity. That openness — from posting long, unedited reads of papers to taking on a leadership vacancy soon after joining DeepMind — helped him move from individual researcher to team lead much earlier than he expected.

Key Points

Nanda struggled with perfectionism and fear of starting projects, worried they might fail.
He practised shipping: writing one blog post a day for a month to break the perfectionist cycle and raise his profile.
Sharing imperfect work (including an unedited three-hour paper read-through) drew tens of thousands of views and attention.
Being visible and saying yes to unfamiliar or scary opportunities expanded his “luck surface area” and led to a leadership role at DeepMind.
His experience underlines that deliberate action and exposure often matter more than waiting for perfect conditions.

Content summary

Nanda admits he often hesitated to begin projects because of the risk of failure. To counter that, he created a constraint — daily blogging for a month — which forced him to produce, iterate and publish quickly. The practice not only helped him gain recognition within the AI community and influence the field of mechanistic interpretability, but also expanded his network and opportunities. Months after joining DeepMind as a researcher in 2023, he stepped up to lead the team when the prior lead stepped down, despite uncertainty about whether he would excel in the role. He frames this path as both creating conditions for luck and choosing to accept challenges when they appear.

Context and relevance

This story sits at the intersection of career development and the fast-moving AI research world. Mechanistic interpretability is a growing subfield important to safety and explainability; rising researchers who build visibility can shape both research agendas and hiring/leadership flows. The piece is relevant to early-career researchers, managers scouting talent, and anyone trying to move from individual contributor to leader in high-skill technical fields. It also reflects a broader trend: in competitive, fast-evolving sectors the ability to ship work and be visible often accelerates career progression as much as technical depth.

Why should I read this?

Quick take: if you habitually over-polish things or wait for a perfect moment, this is for you. Nanda’s hack — force output, say yes, and let the world sort the winners from the imperfect attempts — is a tidy, actionable slice of career advice. It’s not deep management theory, but it’ll nudge you to stop overthinking and start doing.

Author take

Punchy: Useful, short and practical — we’ve saved you the time of listening to the podcast and picked out the parts that actually help you get unstuck.