The 1e⁻¹⁰x Data Scientist.

Inspired by Bojan Tunguz’s post 10x Data Scientist.

Founders: if you ever come across this rare breed of Data Scientists, grab them. If you have a 1e⁻¹⁰x Data Scientist as part of your starting Data Science team, you increase the odds of your startup success significantly.

How do you spot a 1e⁻¹⁰x Data Scientist?

1. e⁻¹⁰x Data Scientists are fast. They start data science projects by training a classifier before any data cleansing or analysis.
2. e⁻¹⁰x Data Scientists are efficient. They write thousands of lines of code without ever defining a single function.
3. e⁻¹⁰x Data Scientists are focused. R is the only language they will ever need to know.
4. e⁻¹⁰x Data Scientists are stylish. In interviews, they describe a programming pattern as a new silicon valley fashion trend.
5. e⁻¹⁰x Data Scientists are web savvy. They’ve only written code in Jupyter Notebooks.
6. e⁻¹⁰x Data Scientists have strong commenting skills. At any given time, at least 85% of their code is commented out, and the rest is commented as not working, revisit later.
7. e⁻¹⁰x Data Scientists are confident and inclusive. They’ve have never needed to write a unit test and in interviews they define TDD as Telecommunications Device for the Deaf.
8. e⁻¹⁰x Data Scientists are historical domain experts. Deep Learning Research Scientist title is on every previous job on their resume, going back to 2003.
9. e⁻¹⁰x Data Scientists are perfectionists. They quickly arrive at models that produce a perfect AUC of 1.00, and proudly show off their work.
10. e⁻¹⁰x Data Scientists are flexible. They don’t use any experiment tracking framework, and just “go with the flow”.
11. e⁻¹⁰x Data Scientists know what they want in life. They only use XGB, regardless of the domain, because it’s the one that won a lot of Kaggle competitions.

Author’s note: this article is a satire, please don’t take offense. Every data scientist is guilty of all of these fallacies at some point in their career and (including, and especially me) we need to laugh at ourselves on occasion. Practice makes perfect, never stop improving your craft, never stop learning.

I strongly recommend the FastAI v3 course part 2 for deep domain knowledge narrated by very excellent, talented educators.

I’m a proponent of functional programming and functional data engineering.

EvolutionIQ is looking for data engineers and data scientists passionate about solving hard problems. Details at evolutioniq.com. If you’re interested in joining the founding team contact me directly on LinkedIn.