Data Science in Consulting. Career Path Guide

Data Science in

Consulting.

Career Path Guide

CONSTANTINE STOKOLOS, MBA

Agenda

• Who is Constantine Stokolos?

• My Career Path

• Data science : Consulting vs Industry

• How to improve? Resources

• QnA

Who is Constantine Stokolos?

• Senior – Machine Learning Engineer - Advisory DnA - EY

• Ex IBMer

• MBA, Georgia Southern

• Master of Data Science ( Applied Statistics ), Kennesaw State

• linkedin.com/in/constantinestokolos/

21/05/2018

Industries :

- CPG

- Retail & Fashion Goods

- Utilities

- Fraud Prevention

- Postal

- Pharma

Sectors :

- Public

- Government

- Commercial

Clients :

- Nike Inc

- Southern Company OpCos

- USPS

- Social Security

Administration

- Pfizer

Career path

21/05/2018

PROGNOS INC

SOUTHERN

COMPANY

IBM EY

2012 2013

2016

2018

SAS

SQL

SAS

SQL

Python

SPSS

SQL

Spark

Scala

SAS

SQL

Python

SAS

Tensorflow

Keras

MxNET

Python

Scala

Spark

SQL

Tensorflow

Keras

MxNET

Data Science : Consulting vs Industry

21/05/2018

Consulting Pros

- Higher Compensation

- Lower technical requirements

- Relatively easy to get in

- Consider yourself a ‘product’

- Faster Growth

- Exposure

- Various industries

- Traveling

Consulting Cons

- Sales. Sales. Sales!

- More Work. ( 45-50 hrs + )

- Client comes first

- Lots of miscellaneous activities that take

your time

- Poor working conditions

- Traveling

Data Science : Consulting vs Industry

21/05/2018

Industry Pros

- More balanced week work hours

- Office space

- Stable environment

- Decisions are not driven by client

- Quality of solutions are usually higher

- More hours on actual development

Industry Cons

- Lower salaries

- Higher technical requirements

- You are an overhead

- Less exposure

- Slower growth

- Same industry, repetitive tasks

Data Science : Consulting vs Industry

21/05/2018

Trends

- Interactive Dashboards & Reporting

Automation

- Big Data processing : Spark

- Deep Learning

- NLP ( Both Deep Learning and Traditional )

- Cloud Computing : AWS, Azure, others

- BI Enterprise solutions

- Fraud and Cyber Security

- Medical : Image Processing, Anomaly

Detection, Pattern Recognition, etc

Data Science : Consulting vs Industry

21/05/2018

My opinion on how to proceed :

- Start with consulting

- Get familiar with various industries

- Get Exposed to lots of companies

- Switch to industry if consulting is not for you, about Manager and up, or 5+ years of experience

- Stay in consulting if Sales is what you like. Rewards are huge!

- Stay away from time consuming tasks that don’t improve your skill set or boost your resume

Real Life Experience. Must Haves in your Skillset

21/05/2018

1. Python :

1. Core : Numpy, pandas, matplotlib, seaborn

2. Machine Learning : Sklearn, SciPy

3. Deep Learning : TensorFlow, Keras, MxNet

2. Spark, SQL:

1. PySpark

2. Spark Shell

3. Spark Mllib

3. R

4. SAS

5. Cloud :

1. AWS : EC2,EMR, Lambda, S3, others

2. Microsoft Azure

3. GPC

6. Hadoop Framework

7. NoSQL & SQL

8. ETL ( example Apache Sqoop )

Vitals :

- Linear Algebra

- Stat Methods

- Probability Theory

- Calculus

- Bash ( Unix Shell )

https://www.linkedin.com/jobs/view/advisor

y-services-manager-data-scientist-at-ey-

483918204/?utm_campaign=google_jobs_ap

ply&utm_source=google_jobs_apply&utm_

medium=organic

Real Life Experience. R, Python, SAS, Others

21/05/2018

- Python, SQL &Spark

- Must have for anything,

especially ML, DL

- R

- when a company tries to

switch from SAS or have

already some R code in the

system

- Scala

- Native Spark API, master if

you want a clean Spark

knowledge

- Java

- Barely used in consulting as

development time is too long

https://www.kdnuggets.com/2017/09/python-vs-r-data-

science-machine-learning.html

R :

- R good for exploratory

- R - powerful statistical modeling

- R doesn’t scale

- R misses some classes and objects

- R No native Deep Learning libraries

SAS:

- Expensive

- Falls behind on availability of new algorithms

- Have their own algorithms engines, black box

- Often you need multiple packages or even software:

- Eminer

- Base / EG

- ETS, HPF, etc

• Ideally you need to be both

• Math background -> Machine Learning Engineers

• Computer Science -> Big Data Architects

• Typically separate positions on a consulting project

• Not so much in the industry

• At the end it is a matter of your preference what you like to do more

21/05/2018

Real Life Experience. Machine Learning vs Big Data

Real Life Experience. Continuing Education

21/05/2018

- A must do!

- Tools change all the time -> need to keep up and adopt and adjust

- Online Resourses :

- https://www.udemy.com/ Udemy

- https://www.coursera.org/ Coursera

- Bootcamps :

- https://nycdatascience.com/data-science-bootcamp/ New York Data Science Academy

- Meetups :

- https://www.meetup.com/Advanced-Spark-and-TensorFlow-Meetup-New-York/events/248201412/

- Books :

- http://www.deeplearningbook.org/ <- FREE

- http://www-bcf.usc.edu/~gareth/ISL/ <- FREE

- https://www.manning.com/books/deep-learning-with-r <- Good book for R and Keras

- Mock interviews & Resume Critique

- https://www.evisors.com/

Your turn!

21/05/2018