Launching the ATD Data Science Accelerator Program

In November, 2018 we introduced the world to our Data & Analytics Addicts at American Tire Distributors. Since that time our Analytics team has gone on to produce an award-winning tool, develop a recruiting strategy for attracting and retaining top analytics talent and fully embraced ATD’s culture of knowledge by providing associates the opportunities to develop their analytics knowledge.

They bring associates together for monthly “donuts & data” sessions and share project overviews through our ATD blog.

But, is that enough? Could they do more?

These are some of the questions that the Analytics team asked themselves. This is how ATD’s Data Science Accelerator (DSA) program was born. The program needed to add value to the business and our associates. In the end, we developed a program that provides associates the opportunity to explore the world of data analytics, work 1:1 with analytics team mentors and the opportunity to be put on a “fast track” interview process for a junior data scientist role, if available.

For 6-months, through an enriched learning environment, associates will learn the basics of Python programming, statistics and machine-learning. With this new-found knowledge and the support of their mentors, each associate will go through a capstone project. The program adds value to the business by providing graduates of the program the skills needed to tackle business problems with data-driven solutions.

Let’s look at the 4-key areas of our DSA program: Content of the Program, Tech Infrastructure, Recruiting Participants and Program Cost. We recommend considering these areas when standing up your own DSA program.

 

 

Content of the Program

The 6-month DSA Program is broken into two segments: online learning and a capstone project. During the first 4 months, participants complete various online learning courses. These courses cover the fundamentals of data science, including the basics of data ETL using SQL, Python programming, Data manipulation, machine-learning, statistics and a few other advanced topics.

Our analytics team put a lot of thought into what the curriculum should be in order to maximize the knowledge output for the participants. The online learning modules take 10 to 15 hours per week to complete, on top of their normal work load. It was critical to have the correct courses and the order of how the participants would progress through the modules.

Another critical area was providing the proper support throughout the program. Meet David, Salman and Sid, the DSA mentors from our analytics team. They’ll provide the support and guidance to participants through weekly mentor meetings and a dedicated slack channel. Each mentor will work closely with a pair of participating associates.

Analytics Team Mentors Left to Right: David, Salman and Sid

After completion of the first 4 months, participants will work with their assigned mentors and complete a 2-month long capstone project. The capstone project will focus on a business problem from their respective business unit and will require the development of an actionable data-driven solution. To showcase their talents and capstone project, we provide participants the opportunity during a demo-day to present their data-driven solution.

Tech Infrastructure

Given the technical requirements of the program, our associates needed the proper computing environment. Immediate challenges we identified:

1. Most associates use laptops that don’t have native support for Python.

2. The laptops range in a variety of models, with different versions of Windows, hardware components and overall computing power.

We asked ourselves: “How do we make a consistent and standardized computing environment for each associate based on these challenges?”

The answer is in the Cloud. The Google Cloud to be exact.

With assistance from our internal Google Cloud support team and IT, we set up a separate Google Cloud Platform (GCP) dedicated to the program. The GCP environment enabled us to host virtual machines (VMs) for each participant in the DSA program.

We recommend using a similar setup as this solution has many advantages:

1. All VMs run Linux Ubuntu operating system, arguably the most Python/Opensource friendly OS out there.

2. Each participant has their own VM, allowing an easy to maintain individual dependency, while keeping hardware level settings consistent

3. VMs are easy to reconfigure, so we’re able to add/reduce compute power, memory and access to different databases across the organization.

The only requirement for associates would be to install the Linux Subsystems on their laptops. They’ll use the Linux Subsystem terminal to access the remote GCP VMs. This allows them to leverage the remote GCP VMs throughout the DSA program for their coursework and capstone projects. Problem solved!

Recruiting Associates

For the inaugural DSA program, we decided to cap participation to 6 associates. We wanted to ensure our mentors could provide each associate the support and guidance needed to succeed. It was important to make the program diverse and fill the 6 spots with associates who represented different business functions, backgrounds and industry experience. After creating collateral that helped inform associates of the DSA program (flyers, FAQs, etc.) and setting up a Google Sheet sign-up website, we introduced the program through ATDs internal communication channels, and our leaders, encouraging associates to apply.

What happened next blew us away! After 2 weeks, more than 120 associates from across the country submitted their application. Applicants represented various business functions and levels throughout ATD. The submitted applications were very impressive and ranged from a delivery driver in Oklahoma to a senior director in finance.

Questions from the DSA program application

As part of the application, each associate was asked to rate their level of expertise with Excel, SQL, Virtual Basic and if they’ve had any exposure to Python or R programming.

The most important section of the application covered “why” the associate wanted to join the program and what they hoped to get from it. This provided the insight into what motivated the them.

Through much deliberation and discussions with business leaders, we selected our very first DSA program participants. Congratulations to Colleen, Daniel, Doug, Mike, Sridhar and Steven on their acceptance into the inaugural DSA program! These associates represent our sales, revenue management, supply chain, finance, IT and eCommerce teams.

ATD Associates Left to Right: David, Doug, Mike, Sid, Sridhar, Colleen and Alex

Program Cost

When we set out to create this program, one thing we were unsure on was if this was going to be cost efficient. After doing a lot of research and coming up with solutions to the challenges we faced, we were able to design a program that costs less per associate than many analytics online certificate programs!

The cost for the entire DSA program includes the online learning licenses, setting up the GCP VMs and we even calculated the cost associated with support from our ATD business partners.

We’re thrilled to give associates the opportunity to learn, grow their knowledge and be able to provide actionable data-driven solutions to their business functions.

We’re excited to launch this program and further embrace ATD’s culture of knowledge. We look forward to sharing future updates on how the program is progressing.