In a few weeks, I’ll cover this pattern in my talk at @DeliveryConf about patterns and pains in delivering data-driven solutions. You should be there.
-
-
Prikaži ovu nit
-
This isn’t to say that you shouldn’t hire PhDs. But there are two things: 1) you don’t need a PhD to do logistic regression or random forest classification, which cover 95% of practical enterprise data science use cases
Prikaži ovu nit -
2) if you do need a PhD, you need to pair them with an equivalently-experienced engineering *team*
Prikaži ovu nit -
That means data engineering, software engineering, infrastructure, and qa support, *at a minimum*
Prikaži ovu nit -
If you want to hire a PhD data scientist to build production features, plan on 3-6 people with them.
Prikaži ovu nit -
But actually, this is true for any level of Data Scientist—because software delivery is a team effort. It’s just that too often “PhD” is intepreted to mean “entire IT department”
Prikaži ovu nit
Kraj razgovora
Novi razgovor -
-
-
Hoo boy. Software engineering is a skill, it turns out
-
Tbh i have seen that with team with delivery experience too. Happens that it is far easier to do "software" if you don't have to deliver anything else thanLOC
- Još 1 odgovor
Novi razgovor -
-
-
“CONSULTING: If you’re not part of the solution, there’s good money to be made helping prolong the problem.” —Despair dot com
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Yep. The anti-pattern is: first hire any junior data scientist (PhD, BS or online courses). Been covering it in talks for 4 years now :-/. The situation only marginally improved now, primarily because bigger cos realized they screwed up and rebuilt that area (not always well).
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.