WELCOME to our Mentoring Program in Data-Science! 

84% of the CEOs think that Data-Science & AI will be required to reach their growth target. 

This figure is one of the main conclusion of a survey done by Accenture last November, polling 1,500 CEOs worldwide. 

¾ of these CEOs think that if they don’t develop these skills internally, they may well be out of business within 5 years. 

But,  76% of them admit that so far, they fail to scale their AI/Machine Learning initiatives. 

Why is it so difficult to scale on Data-Science?

Data-science (whether we talk about Analytics | AI | Machine Learning |Mathematical modeling..) is somehow new in many industries. Not that these techniques didn’t exist before, but they were mostly used in academic research and in some advanced fields. 

Now, as data are plentiful, we need data-science everywhere! 

We need Data-Science: 

• In order to make better decisions (so that we no longer rely on gut instinct, anecdotal evidence, human biases and in many cases.. the flip of a coin) 
• In order to create smarter products  
• In order to automate manual processes in service centers like Societe Generale European Business Services for instance 

Problem: experienced data-scientists are a scarce resource .. and once their Proof of Concept is done, deployment in production (according to I.T standards) is hard to achieve.  

2 elements to solve this challenge: 

1 – A centralized platform for Machine Learning  is required to easily check models once they are trained, and push them into production.  We will demo how such a platform could work using open-source solutions within a webinar we host during DevTalks on 12/06 at 14h40 – JOIN US! 

2 – But besides that we need to help people become at ease with data-science so that they can progressively handle projects on their own. We need to train them so that they embrace this  

Back to school in order to learn data-science 

The volume of available online courses is just enormous! Beginner, Intermediate or Advanced levels ? pick your desired level 

Unfortunately, it doesn’t work this way –  The reality is that it is very difficult to really take off on data-science without a sustain effort in time. People will give up, people will drop out – massively. 

Solution: Be part of a community. Do some hands-on projects.. and get some mentoring. 

From beginner to intermediate: Enter a mentoring program 

How such a mentoring program works: 

• Data (and knowledge generated from these data) stay with the learner – it avoids some frequent limitation with data confidentiality  
• Learner can work with a tier-party software solution (no code, just clicks) 
• Or Mentor can provide some generic & open-source code (SG EBS’ favorite solution) which will be adapted by the Learner to the specific use-case 
• Learner commits to work part-time, progressively, on this project (Validation from his/her manager) 
• Project doesn’t follow strict time-boxing as in Scrum/Sprint Planning  - it is a learning effort, not a new feature within a WebApp 
• Instead of talking about “Sprint”, we prefer the terminology of “Review of Progress” with Mentor 
• Learner sends minutes of this weekly Review of Progress to his/her mentor and N+1 Manager = Lean Project Management = no budgeting 
• Best Effort mode – Learners may need to pause – and it’s ok 

Within Societe Generale European Business Services, we have already several people who entered this program. And they also transitioned in their work.  They still have the job title of Business Analyst, Team Lead (SME), Front-End Web Developer Java Developer, Project Manager.. etc.  

Within our Data Community, as mentors – we welcome everybody with a solid enthusiasm for data.  We support women in data-science to develop this vibrant community. 

All in, Mentoring data enthusiasts & setting up a centralized Data-Science platform are the 2 essential axes to succeed in this transformation. 

Within this survey from Accenture, the most advanced companies report a Return on Investment (ROI) x3 on individual data-science projects once this scaling is achieved. 

Nicolas Boitout, PhD & Radu Lupu, Professor 

CoE AAA – AI, Analytics and Automation 

SG EBS