1 of 29

Managing Junior Data Scientists in the Time of Coronavirus

Joel Grus

@joelgrus

Principal Engineer, Machine Learning

title: Managing Junior Data Scientists in the Time of Coronavirus

abstract: Even when they're sitting next to you, junior engineers and data scientists can need a lot of attention and guidance. When they and you are trapped at home, maintaining that relationship can be more challenging. I'll talk about the needs of junior team members, the needs of managers of junior team members, and tips and processes for distributed collaboration when everyone is trapped at home.

bio: Joel Grus is principal engineer at Capital Group, where he leads a small team focused on research, design, and implementation of machine learning and data products for the investment group. Previously he worked at the Allen Institute for AI, where he was a core engineer on the AllenNLP deep learning library; at Google, where he did boring software engineering things; and at a wide variety of startups where he did a wide variety of data science and proto-data-science. He wrote the beloved book "Data Science from Scratch", the somewhat amusing blog post "Fizz Buzz in Tensorflow", and the polarizing JupyterCon talk "I Don't Like Notebooks". He is also the co-host of the Adversarial Learning podcast. He lives in Seattle.

2 of 29

About Me

3 of 29

Investment Group Tech

Investment company since 1931 (American Funds)
Long-term investments based on deep, fundamental research
>$1.7T assets under management

New team within CG started by former AMZN, GOOG, MSFT, and more
Goal: apply new technologies to long-term fundamental investing
We're always hiring (LA, Seattle, SF, London), talk to me

4 of 29

Managing Junior Data Scientists in the Time of Coronavirus

5 of 29

The

Basics

6 of 29

Encourage them to get food

7 of 29

Remind them to wear masks

8 of 29

Instruct them in

(data science poem from https://www.kdnuggets.com/2019/05/data-science-poem.html)

9 of 29

Encourage Social Distancing

Pick Your Lockdown House

House #1

pandas

tensorflow

scrapy

matplotlib

House #2

dask

seaborn

sklearn

jupyter

House #3

keras

spacy

data.table

plotly

House #4

pytorch

statsmodels

ggplot2

nltk

House #5

scipy

streamlit

dplyr

beautifulsoup

House #6

numpy

xgboost

shiny

spark

10 of 29

Remoteness

11 of 29

Telepresence

12 of 29

Omnipresence

13 of 29

Slack

(or some knockoff)

14 of 29

Collaboration

15 of 29

Writing Clean, Documented Code

16 of 29

Robust Test Coverage

17 of 29

Thorough, Constructive Code Reviews

18 of 29

WikiWikiWebification

19 of 29

Autonomy

20 of 29

The Data Science Venn Diagram

math + stats

hacking skills

empathy

substantive expertise

data science

21 of 29

Give them space to explore their reason for being

22 of 29

Put them in charge

23 of 29

Emerging

from

Quarantine

24 of 29

Back to Normal?

25 of 29

Keep it casual

26 of 29

Stay WFH-

indulgent

27 of 29

Help them adjust to their new freedoms

28 of 29

"But Joel, aren't these all the exact same things you've been telling us to do for years even way before Coronavirus?"

29 of 29

Thanks for attending!

will tweet out slides @joelgrus

or find them at https://tinyurl.com/coronavirus-data-science

check out my never-updated blog at joelgrus.com

or the podcast at adversariallearning.com