I'm a DS at Twitch.
I have 6 YOE. My peers in other companies are managers, while I'm still an IC. Twitch is a shitshow for data science, so I'm not moving up.
Sr, Lead, and Staff positions elsewhere seem to require specific domain knowledge, like fraud or healthcare.
For advanced DS out there: how did you transition to higher roles while changing jobs and/or domains? Is domain knowledge a "nice to have," and if so, how do I sell that? Seems like the technical and personal skills are harder to teach, but I don't know if hiring managers see it that way.
If you've wrestled with this problem, please share.
I'm a DS at Twitch.
- Amazon / Data JohnnyUtaSo I tried data science and just wasn’t that good with it. What was good at, is turning business use cases in solvable and deliverable products using data science.
Ever think about Product Management? It’s interesting because many PMs know a little about data science but still try to build unrealistic products. I’ve found a really cool niche and you should try it out.
Don’t even ask or try to get that title. Just literally run point on project from the perspective of a product manager. If it’s something you like, look to make a change.
Best of luck. Have fun and be safe.
- The matter with domain knowledge is that pure stats will bring some results only in the field where they were not used before. After the initial progress stats stop working. At this point you’ll need to dig into the domain and become the domain expert which you probably DONOT want. You must stop being statistician, in a way, sell your soul in order to advance.
- Mathematical statistics is a formal framework that applies under certain assumptions. If it "stops working" that's only because one isn't using the appropriate tools correctly.
On the other hand, domain knowledge has value because it provides understanding regarding the nature of data and points at impactful, meaningful questions to work toward.Oct 13 2
- I mean by stops working the diminished returns. Sorry for not being clear.
Case in point is finance and investing: try making money with data science in equity markets. The stats have been applied here forever, so you need to bury yourself into nitty gritty to squeeze out any profit through math
- There's a downside. Domain knowledge pigeon hole in terms of opportunities. Is that good?
I mean if you are pigeon holed in cloud and ML that's great. But for others like me, I'm working with let's say a traditionally poo-poo part of tech seen as a cost center... Do I want to be king of ML in a shitty field?
- Vrbo MJon50What’s the difference between data scientist and applied scientist at twitch?