Founded in 2015, Moglix, the industrial unicorn of B2B e-commerce, provides solutions to more than 500,000 MSMEs.
Analytics India Magazine contacted Sandeep Goel, Senior Vice President, Strategy and Operations, Moglix. A graduate of the Indian Institute of Science (IISc), he is responsible for designing growth strategies at Moglix. Sandeep has over 25 years of experience in the tech industry.
âToday, each organization is developing specific solutions to its needs. Still, portable models will become available over a period when models developed by one organization will be easily reusable by others, accelerating the deployment of AI / ML solutions, âsaid Sandeep.
GOAL: How is your business implementing AI / ML / Analytics to drive growth? Chat briefly, citing customer success stories or use cases.
Sandeep: TO Moglix, we use artificial intelligence and machine learning algorithms for the following purposes:
- Data cleaning: A clean catalog is vital for our business. Situations such as missing attributes, duplicate items or misclassification can negatively impact our business. AI / ML helps us keep our catalog clean.
- Look for: B2C has led the evolution of search use cases, where the consumer prefers to search for a product rather than browse through various categories. The exact use case is starting to apply in the B2B world, and search results need to be precise. Given the linguistic diversity in India, where English has also become Hinglish (a mixture of Hindi and English), AI / ML becomes a powerful tool for running Natural Language Processing (NLP) algorithms. in order to generate accurate search results.
- Decision making: Data is the new raw material and AI / ML is the mining machine to extract value. Data-driven decisions are essential for improved efficiency, cost optimization, and governance in B2B commerce scenarios.
- Improved efficiency: From mapping customer requirements to our product to predicting potential incorrect addresses, AI / ML helps us make several decisions. We are able to achieve 4 times the efficiency and detect incorrect addresses in advance to avoid product return.
- Cost optimization: The order processing flow involves different decision points, such as the selection of a supplier or a logistics service provider, impacting the cost of the transaction. We use AI / ML to make such decisions, which helps optimize costs.
- Better governance: B2B business transactions are vulnerable to human error, both intentional and unintentional. Techniques such as optical character recognition (OCR) help us avoid such errors to a large extent.
OBJECTIVE: What percentage of the resources are devoted to the implementation of the state of the art?
Sandeep: At Moglix, we focus on creating products and solutions for our internal consumption and our customers. Our experience in supply chain digitization, master data management, data mining and purchasing analytics is at the heart of our business.
Currently 10% of our technical staff is dedicated to working on AI / ML related projects.
GOAL: What kind of skills are you looking for when hiring data scientists / ML engineers?
Sandeep: There are two key aspects of a data scientist:
- The first is the ability to understand and interpret data. This skill usually comes from experience in the field, and one must spend enough time in the industry to acquire this skill.
- Understanding of AI / ML tools and algorithms. Colleges and online learning platforms offer various courses.
We are looking to hire people who have expertise in the above two areas. We love early problem solvers, and few of our data scientists are local.
AIM: What is the major stake in recruiting Indian talents?
Sandeep: The first challenge is the availability of the right talent. The Indian tech vertical has shifted from a service mindset to a product mindset. However, the transition did not go completely and people often confuse the role of a program manager versus a product manager. Indian tech companies need product managers who can take ownership of the product from start to finish, from conceptualization to commissioning. While there are many institutes in the market offering data science course, most of them are not equipped to provide accurate data science experience due to lack of data and proper scenarios.
The second challenge is the stability which deteriorated during the pandemic. Working anywhere has impacted the stability of resources, and the war for talent has triggered attrition driven by salary increases that are not sustainable.
OBJECTIVE: What are the most used programming languages ââby your data science team? What does your team’s tool stack look like?
Sandeep: Some of the programming languages ââused primarily by our data science / analysis team are Java, Scala, PHP, Python, Angular.
Our tool stack includes Scikit, SpaCy, RabbitMQ, Cassandra, Mongo, Elasticsearch, and Apache. Other libraries include Pandas, Numpy, FBProphet, NLTK, SciPy, PDFminer, Pytesseract, APScheduler. For the management of the database, we use an SQL server. Additionally, we use Matplotlib, Seaborn, Plotly for visualization and GitHub for tracking.
OBJECTIVE: What types of AI / ML deployment challenges is your team facing? Do you prefer to have your R&D team in-house or to outsource innovation?
Sandeep: The right amount of good-quality training data is essential for data scientists to examine and build models, whether it’s labeled or unlabeled data or research journals. As such, integrating technology into the manufacturing supply chain ecosystem is rare. An application with AI and ML capabilities feeds on user behavior data to continuously improve. Lack of user adoption weakens the feedback loop between application and user.
Organizations in the pilot stages of purgatory face the challenges of distributing data across silos, teams and functional departments, and users’ personal devices. The integration of disconnected data fragments is a challenge. Other companies have invested in a variety of legacy systems and must navigate between multiple solutions to complete the end-to-end cycle of a task. Again, approval workflows and data governance models are, in many cases, not clearly defined. This gives rise to multiple scenarios of impure data on their supplier base, customer base, warehouse management systems, and inventory management systems.
Innovation can never be outsourced. It must be managed by an internal team with in-depth knowledge of our business. However, we engage with partners to co-innovate and learn from their experiences.
AIM: How do you see the AI ââ/ ML landscape evolving in India in relation to your field?
Sandeep: India’s manufacturing sector will soon hit the $ 1 trillion mark, and faster decision-making capabilities will be needed to manage this growth. The data explosion caused by this growth will be impossible for humans to take, and they will have to familiarize themselves with the technology. AI and ML will play a vital role in creating efficiency, governance and decision making.
Areas such as NLP-based research, deep learning based image processing solutions or prediction and classification algorithms will become an integral part of B2B manufacturing and commerce. Today, each organization is developing specific solutions to its needs. Nonetheless, portable models will become available over a period when models developed by one organization will be easily reusable by others, accelerating the deployment of AI / ML solutions.
Join our Telegram group. Be part of an engaging online community. Join here.
Subscribe to our newsletter
Receive the latest updates and relevant offers by sharing your email.