Machine Learning and Deep Learning are a very sought after skills today, both globally and in India alike. Evolution strategies, reinforcement learning, as well as imitation learning are also gathering momentum in the research communities, and will soon find their way into commercial applications. India will soon witness a lot of action in the AI space, including the development of specific applications and tools in new and emerging areas.
Headquartered in Delhi, this startup Nube Technologies essentially creates value by furnishing Master Data Management (MDM) solutions, powered with AI. The organization was founded by Sonal Goyal in 2010 as a niche big data consulting firm with strong focus on analytics over big data in the cloud – Apache Hadoop, Apache Spark, Amazon Web Services, ML.
Nube helps in managing business data, cleansing it, and providing a comprehensive view to each entity of interest. Analytics India Magazine got in touch with Goyal to know the unique story about Nube Technologies in the AI-driven data management space.
Businesses globally struggle with merging data, from external sources with in-house data, particularly when the in-house data itself reflects multiple copies with variations. Moreover, the tools that exist today lack the scalability and cross domain applicability, besides being deficit in sheer accuracy and performance.
The comprehensive MDM tool — Reifier
The startup has followed an impressive journey in this space, transitioning from consulting to a product development organization. Nube has come across leading practitioners in the space, and received great acknowledgements from them.. Sonal Goyal, Founder, Nube Technologies extols, “We picked up one of the toughest problems in computer science worldwide, entity resolution and record linkage. We are happy and proud to have cracked it very elegantly. It has been a terrific journey.”
Reifier discovers hidden patterns in variations across different entities, to consolidate and deduplicate them. This helps in providing the end-user with one single golden truth. The greatest challenge in entity matching is the fact that it is a computationally intensive problem, i.e. if you have n records, you must match against every other n-1 records, as there are no unique keys to fall back on. So, it can be considered as a scalability issue.
“Defining rules for each entity is tough, and then perceiving these rules change, as the business changes and grows, can be outlined as an accuracy problem,” exclaims Goyal. Enterprises have to put effort on a regular basis to keep pace with changing data. Businesses must deal with distinct kinds of entities, customers, vendors, products, and more, daily. This surfaces as a much bigger business challenge.
Reifier learns from the data as well as makes smart comparisons through multiple proprietary algorithms. So, the tool can handle just any kind of data at any scale with zero customization. Reifier is so good at this pattern discovery that it can handle different languages as well – Hindi, Japanese, Chinese, Thai, etc. with the same ease, and with the same accuracy as English.
Features of Reifier:
- Reifier seamlessly scales from a few thousand to millions of records, in real-time. This tool is incredibly fast.
- Reifier’s AI engine works with the same ease for data variety in multiple languages – business names, addresses, customer names, educational institutes, cameras, mobiles, watches, other products.
- No manual configuration of rules and algorithms is needed. This results in lightning fast deployment and unmatched versatility and accuracy.
A tightly regulated industry like insurance has very stringent compliance reporting. However, most insurers have a decentralized IT infrastructure, where different business lines have different systems. This is efficient for the business, focusing on the BU and specific agent or broker needs.
Each line of businesses, viz., home, automotive, commercial, and surety are maintaining their own systems. Unfortunately, data is often duplicated within these systems. Reifier consolidates this data for businesses, providing them with clear opportunities for cross selling, risk assessment, and compliance reporting. Reifier also checks their internal customer information against externally sourced data, to ensure that sales outreach happen to the right people.
Nube Technologies — adding value to enterprises
For any business to function properly, there is a clear need for a view of the customers they are servicing, the products they are buying or selling, the vendors they are engaged with, and so on. Goyal remarks, “Businesses need to have complete knowledge about all business touch points and entities.” However, this information is spread across different business units, in myriad applications across the enterprise.
Captured data about entities has variations like missing fields, typographical errors, and format issues which makes consolidation a big challenge. Moreover, the growing data sizes adds another dimension to the exploding problem. This is where Nube comes in, providing better sales and marketing decisions, and better analytics and risk compliance, throughout all the data.
Nube leverages various techniques to ensure that their solution solves the customer pain point. The approach has always been to define the business problem first, and then findingg the best way to solve it technically, while accounting for scalability, performance, accuracy, and ease of use. “We also look at making the base platform cross domain, so that we can address a sizable market,” comments Goyal.
Life at Nube Technologies
Goyal shares, “The team here is exceptional and well aligned with the company ideals, reflecting strong technical skills, deep domain expertise, and relentless product focus.”
“We are very picky about what we work on, it has to excite us. The work has to be valuable to the end user and we should be able to add a lot of value to the problem,” remarks Goyal. The company believes in innovation, and the team here is bound together by dedication for the customer and the joy in solving tough problems.
Nube is undertaking couple of interesting projects around enterprise mobility. Some of these upcoming projects include model feedback, building a single copy by merging merged records, and support for more data sources, and GUI enhancements. Nube has also noticed a lot of action on the sales side in the recent past, which has prompted the firm to streamline sales-related activities.
However, the company will continue to focus and innovate upon its primary offering – Reifier. “The organization has tuned the Reifier engine to accommodate disparate dataset from millions of records. This has brought to lights some phenomenal results,” concludes Goyal.
Try deep learning using MATLAB