What Five Years at a Startup Taught Me

I joined a startup five years ago as a data scientist. Most of what I believed on day one turned out to be wrong, or at least incomplete. This is the list I wish someone had handed me at the start — in three parts: the work, the career around the work, and the life around the career.

On doing the work

1. Business impact is the only scoreboard. Roles are building blocks toward a goal — the goal doesn't care which one you are.
2. Don't chase fancy methodology — ship fast. The biggest error source is building the wrong thing; shipping is how you find out cheaply.
3. Allow yourself to fail. Failure is the only experiment with a proper control group. Fail small, do the post-mortem.
4. Be proactive, opinionated, and loud about it. Falsifiable opinions turn you from a service function into a thought partner.
5. Analysis doesn't exist until it's communicated. The narrative isn't packaging around the work — it's the last mile of the work.
6. Take ownership — everything is learnable. No skill sits behind a wall. Own the outcome, including the parts you can't do yet.
7. Set goals and milestones for yourself. Define your own successful quarter, or the ambient chaos defines it for you.
8. Work smart, not just hard. Impact = direction × effort — and direction can be zero. Point before you push.

On building a career

9. Senior roles come from connections, not cold applications. Do visible work, write publicly, stay current — be the person who comes to mind.
10. Reputation compounds like everything else. Built in small deposits, spent in one overclaimed result. People refer who they trust.
11. Past a point, your impact is what you enable. Making five people better outships anything one person can do alone.
12. Keep learning, permanently. A standing line item the job is never allowed to fully crowd out.
13. AI evolves — statistics is the moat. Tools churn every six months; the statistical foundation rusts too. Brush it deliberately, often.

On everything else

14. Keep a regular workout. One of the few domains where input reliably produces output. It keeps you sane.
15. Invest early — buy experiences, not luxury goods. Time in beats timing. Experiences appreciate in memory; objects depreciate in a drawer.
16. There is no balance at the frontier — and that's a choice. Outsized goals take outsized dedication. Choose it deliberately, for a bounded season.

The compressed version

Move the business, not the methodology. Ship, fail, own it, say what you think — and make sure it lands. Point the effort before you spend it, then multiply it through other people. Let your network, reputation, knowledge, health, and money compound — they run on the same math. Chase the frontier, but keep the statistical foundations polished — that's the moat. And if you choose intensity, choose it on purpose.

Five years in, that's the whole list. Ask me again in five more.