Data Science Management

A data science team still has to hire, interview, write, and stay legal. The page follows that order: recruiting and interview questions first, then the politics and design docs of the daily work, then contracts and due diligence, and finally general notes on keeping data scientists effective.

Hiring / recruiting

A team starts with a hire, and the first thing to settle is which skills you are hiring for. Data engineer skills is a brief data engineering interview prep guide, written by a data engineer of three years who noticed a distinct lack of readily available resources, partly because the role varies from company to company and domain to domain. The skills it expects are:

1. Coding (Typically Python)

2. SQL

3. Database design

4. Data architecture/big data technologies

5. Soft skills

Interview questions

Once the skills are named, the interview has to test them. These question banks cover ensembles, trees, and k-NNs.

The same notes are in Decision Trees, Ensembles, and KNN.

40 questions on ensembles is Ankit Gupta's set of questions on ensemble modeling techniques to test a data scientist, covering bagging, boosting, stacking, and voting. 30 on trees is his companion set on tree based models, including decision trees, random forest, and boosting algorithms. 30 on knns is Sunil Ray's KNN interview questions for freshers and mid-senior professionals, with explanations and solutions.

Politics

Interviews are not the hard part once the team is in place. The most difficult thing in ds, politics is the post "The most difficult thing in data science: politics", originally published on rDisorder.

Writing docs

Politics is easier when decisions are written down. Design docs at google describes design docs as one of the key elements of how Google builds software.

Legal & contracts

Docs are not contracts, and advisors need one. (FAST) Advisory board saas agreement is the FAST Agreement, used by tens of thousands of entrepreneurs and advisors per year to establish productive working relationships, trading advice and support for a standardized amount of equity.

Due Diligence

Contracts sit beside diligence, where someone outside checks whether the ML is real. The guide by Inbal Budowski Tal is how to run a machine learning technical due diligence for VCs, estimating the depth of a startup's ML technology and the team's capability to deliver their "claim" and win the "fame".

General

After hiring, docs, and contracts, the last question is how to keep data scientists effective once they are there.

1. The Care and Feeding of Data Scientists - O'Reilly is the O'Reilly report on that question.

Transitioning From AI Teams To AI Squads (October 2024) is the team shape on this page. Setting Up Data Science Teams For Success (October 2022) is the team-setup note, built around Dr. Ori Cohen's three pillars of data science, and the data-team page is the engineering side of the same idea.

Deprecated links

These links and images no longer work. The original wording is kept here. A same-resource copy, when one was checked, is used above.

• The most difficult thing in ds, politics. This address no longer opens: https://towardsdatascience.com/the-most-difficult-thing-in-data-science-politics-ca6dd781da56