[Case Study] Maximizing SWE Growth: Ownership and Specialization At Startups

**Huge** thanks to [Henry Prior](https://www.linkedin.com/in/henryprior/) for joining us and sharing his perspective and insights!

This was a *massive* session, so here are just some of the highlights:
- **Henry's core learning**: Bias towards action. You aren't really going to truly understand what is and what isn't fruitful to you until you just start trying things.
- **Story #1** - At his first startup job, Henry found a project to upgrade from Tensorflow 1 to 2. This ended up moving the metric he cared about, but he found out later that the business impact wasn't great. The company's clients were non-technical and didn't really care about the company's model being ultra-accurate.
- **Story #2** - Found a reinforcement learning project at his next company where he did all the modeling on his own and worked closely with an infra engineer to ship. The work was very complex and had huge impact, saving ~25 hours per week in Data Scientist cycles. He got massive recognition from the company, and this project was a big part of his move to senior.
- You have a lot of autonomy in early stage startups. Take advantage of this and actively look for large problems to solve, doing so as early as the interview process. It's not like Big Tech where you are generally given a good amount of direction on what to do (at least until you get to staff).
- *It's in your best interest to be completely authentic when interviewing with a startup.* It's not like Big Tech where you're putting on a facade to meet their interview requirements. A startup might not be the right for you due to a lack of work in your interest area: Big Tech in contrast has a little bit of everything.
- *If you're going to get into machine learning, you should understand the underlying math first.* This puts you in a much better position to work on the modeling part of ML vs. the more tertiary tasks like data wrangling. The math is a combination of probability/statistics, linear algebra, and calculus.
- **You can't study your way to being good at system design.** Resources like books and blog posts help, but system design is something you very much have to get good at by doing. Instead of just leaping into the code, start planning out your tasks and doing technical writing. From there, share your material with your team and get feedback.

Related resources:
- ["What's the difference between Big Tech and startups in terms of learning and growth?"](https://link.jointaro.com/yzWgzTF4WFzfXjS29)
- ["Does it make sense for me to choose a specialization like ML? If so, what's the impact?"](https://link.jointaro.com/jV34Jz7cJqQv2MpD8)

Huge thanks to Henry Prior for joining us and sharing his perspective and insights!

This was a massive session, so here