Haryana based start-up, Ecom Express, is one of India’s leading end-to-end logistics providers in India’s e-commerce industry. In the past few years, the company has invested heavily in leveraging emerging technologies for automating their delivery and network optimisation solutions. Leading this digital transformation is Dr Bhupinder Singh, Head of Data Science at Ecom Express. He joined the company with a Ph D in Mechanical Engineering and patents in data science. In an interview with Dr Singh, Analytics India Magazine explored his journey in data science and the company’s digital transformation process. 

AIM: Tell us about Ecom Express’ technological offerings.

We are one of the largest technology-enabled end-to-end logistics providers in the industry today. Our wide footprint reaches the remotest parts of the country that very few other players can do. We’ve been investing in technologies that help us build the network’s right speed and make the overall user experience more seamless. We are making our applications smarter by bringing more automation into our process and digitally transforming them. Embedding models into our technology stack and investing in hardware have given us a huge advantage in processing shipments with speed. 

Over the last few years, we have invested heavily in data science. Our business continues to drive better automation and smarter decision making through data science. Now, we’ve invested in some of the core areas of logistics. For example, when processing many shipments, you need to know that your facilities are positioned right – that they’re closer to the demand centre and have high throughput. This decision making happens through data science. We have one of the largest networks in the country with the widest presence and are especially strong in reaching the remote parts of the country. And as we continue to push more volumes in that, a good percentage of that growth is through the three-four-tier cities, and we’re using data science to position some of our facilities in the last mile optimally. In terms of automation, we are trying to automate our decision making. For example, we are embedding AI into our performance management. Now, AI tells our organisation what benchmark performance should have been in a given geography. 

We’re also using data science for automation and routing. For example, we’ve automated our full last-mile routing systems that help improve overall productivity. Instead of the human, you’re letting the AI decide how many shipments to allocate in a particular route. As a result, it improves the overall productivity in our last mile and the midway.

AIM: Can you expand on the tech stack behind your platform?

The first mile of operations is picking up shipments from customers and manufacturing warehouses, processing, and getting them ready for a third-party logistics provider like us to come in. We bring our large scale robotic automation systems that allow the processing of these shipments at a very high speed. 

The mid mile operations transport these shipments in large trucks from one large facility through a network of facilities to the final destination hub and then to a distribution centre. The last mile is a biker going out and delivering shipments from our distribution centre to someone’s doorstep. These transactions happen through the shipment’s lifecycle and our core technology stack, the ERP, handles it. It’s called Caddy and forms the backbone of everything that we do. It stitches together all the data through multiple touchpoints as the shipment moves forward. We have integrated our HRM system with third-party vendors. A data push and pull happens through our APIs onto our core ERP. We also have data science APIs that integrate microservices into our tech stack. Because of this microservices-based architecture, we get a lot of flexibility for seamless integration with different players, vendors and customers. We can integrate our data science models into our IT applications seamlessly and make them intelligent enough to make automated decisions. 

AIM: Some platforms providing delivery services have come under scrutiny for their harsh algorithms on delivery personnel. How does Ecom Express ensure fair and safe algorithms for delivery?

Quick commerce is said to grow to almost 5 billion in the next five years. And there’s an increasing hunger to cater to the customers with speed. What happens in the background is very complex. You have to start working towards catering to the demand before it has even arisen. For instance, you probably start to cook the biryani before someone has even ordered it in the food delivery space. The background is a lot of analysis and data science that predicts the pockets where the demand will come up. Another significant aspect involves technology and infrastructure to optimally position your facilities and bring them closer to your demand. Earlier, if someone had ordered a laptop in Bangalore, it would have come from Delhi, and you would have waited for a few days of manufacturing. Now you are bringing your stocking closer to the demand centre, where our fulfilment centres are. Hyperlocal centres have also started to come in to cater to the demand. Algorithms help by intaking historical trends of productivity and travel times to help with more pragmatic routing decisions that don’t compromise the safety of individuals who are the field staff delivering to the doorsteps. The safety of field executives is of paramount intelligence. So the debate is very relevant.

AIM: What is the technological culture like at Ecom Express?

We have four verticals in our company today that enable data science. We have teams specialising in data engineering, large-scale data processing, and building the foundational data architecture. This is all on the cloud, so we can stitch the entire view of the shipping journey and bring that to our business users to automate and fasten our decisions. We have a vertical on AI where we have specialists who bring in their skills in geospatial analytics, which is key in a highly distributed environment like ours. We have people who specialise in computer vision and image processing algorithms. We also have people with expertise in deep learning and natural language processing to process data sets, images, text or structured data to build intelligent applications. Finally, we’ve had the vision of embedding data science in every single function of our company to digitally transform various processes.

Our team of business analysts leverages data to bring business insights. After starting the data journey, you start to get a lot of feedback, starting your digital transformation journey. You have to start to learn from the feedback coming in. You have to start to work with it to digitally transform the processes. That change management is a very crucial part of enabling value to unfold from data science, and that’s what we do together as a team, having so many skill sets under one roof, working on data science.

AIM: Tell us about your journey in data science. What made you pursue the subject?

I started my journey as a mechanical engineer back in 2001. But, I was coding when I was in the fourth grade. And by the time I finished college, I was into C++. By the time I finished my Ph D, I had worked to do patent analytics, giving technological consultancy to different inventive labs across the globe in GE. And I was playing with large amounts of text data. By then, I’d seen the journey that technology had taken. Back in 1995, Windows 95 was there, and by the time I had graduated, I was at GE. The processor speeds had grown by leaps and bounds. You could now take in a lot more data than was previously possible. I saw that as an opportunity. I saw how data science could transform so many sectors at a rapid pace; it naturally drew me to data science. And with my background in Ph D, I had exposure to advanced mathematics. I had the exposure to programming to take this on. And I don’t regret that decision at all.

AIM: As a Ph D graduate, what are your thoughts on the importance of internships and academic education in building a successful data science career?

Having a lot of cross-functional experience from diverse backgrounds is key. That is how we hire at Ecom Express. We have a strong collective experience within the team. We have had people who come in with spatial experience, computer vision, image processing, and people from MBAs to people who are computer scientists with specialisations. We have people who have specialised in biomedical; we have people who specialise in geoinformatics, mechanical engineers and industrial engineers. For example, many industrial engineers and production engineers do a great job in our area because they have a lot of exposure to operational research and solving large scale linear programming or optimisation problems and simulation. Such diversity comes together to make a strong data science team.

AIM: What are the essential skills every data scientist should have?

At Ecom Express, we look at data scientists with exposure to advanced math with experience in solving multiple case studies or complex business problems, so they have the right business acumen. We look for people who have strong programming experience because data science is not just building an AI model but also putting it into production. Skills in cloud are extremely valuable. And our tech stack is mostly on Python, so we look for extremely strong people in Python with experience in big data. We have a lot of people who come in with specialist skills in running large scale optimisation algorithms. For example, there are technologies like CBL to solve optimisation problems. Metaheuristic algorithms are very complex and very relevant in our space. So we have people who work on them as well.

AIM: Tell us about Ecom Express’ plans.

These are exciting times for our company and the industry in general. The eCommerce industry is valued at 40 billion, and it will grow to 140,000,000,000 by 2025. In our company, we are helping that growth by building a robust technology stack in the background and investing in advanced robotic automation and other technologies, helping us speed up our entire operations. We are investing in our fulfilment centres so that they are closer to the demand centres and then will help faster lifecycle closure for the commerce industry. We are also increasingly leveraging data science to digitally transform our processes. We will be entering a space in the near time where you will see many more connected supply chains. As 5G kicks in and you see a proliferation of IoT devices, you will be collecting data at a far higher velocity, and you have to be able to process all of that high-velocity data at scale on the cloud through distributed computing. We are investing heavily on that front. We are working towards near real-time decision making in the short term, and all of this together will, I think, help us grow in traditional eCommerce and quickly.