Despite best intentions and mountains of data, organisations often struggle to offer seamless journeys. This week’s start-up – Scienaptic Systems – claims that its unique analytical framework can infuse intelligence in the way organisations decipher customer behaviour patterns, design intelligent interventions and create real value in delivering a unified customer experience.
Scienaptic, a new age analytics company offers end-to-end decisioning solutions powered by a Machine Learning (ML) and Artificial Intelligence (AI) based platform — Ether. Headquartered in New York, Scienaptic was started in 2014 by Pankaj Kulshreshtha and has an engineering centre in Bengaluru.
Analytics India Magazine spoke to Kulshreshtha, who gave us deeper insights on the company, the growth story as well as the future of the company.
While the company offers solutions in Credit Underwriting, Model Risk Management, Predictive Journeys, Fraud Management, Debt Collections & Analysis etc, its main product is Ether. “Ether is designed to work with any of the client’s existing big data or relational database infrastructure. Ether is well integrated with Hadoop and Spark and is designed to operate prescriptive solutions such as Credit Underwriting & Credit Line Management, Fraud Prevention, Model Risk Management, Predictive Journeys, Collections, etc,” Kulshreshtha said.
“These modules can run predictive models, transform data and create features from both structured and unstructured data using Spark Processing Engine and Spark ML library as well as build visualizations and dashboards. Additionally, the models can be exported in PMML format for implementation at the client’s site, or use the Ether scoring module to implement the models built,” he added.
According to Kulshreshtha, Ether has built-in complex data transformation abilities, which helps it address data quality issues. Scienaptic claims to have developed AI within Ether, which points out inconsistencies in the data set without even importing it. Moreover, it has the ability to solve complex data quality issues for both continuous and categorical data with a simple click. Smart transformations ensure that the user need not spend time analysing every table, join keys or structure. Ether provides key concept maps eliminating the noise in a seamless manner. The data undergoes missing value treatment, one-hot encoding for categorical features and capping of outliers in Ether, among other transformations.
The other deliverables:
Apart from its solutions on Model Risk Management, Collections Analysis & Predictive Journeys, Scienaptic has use cases on Credit Underwriting & Fraud Management, fully built out on its platform that delivers full-fledged processes with advanced Machine Learning. Kulshreshtha gives us some examples of deliverables pertaining to these use cases:
The company implemented a smart credit underwriting & credit management platform for a leading credit card company in 3 weeks by leveraging advanced Machine Learning techniques. The solution offered complete model explainability and adverse action reporting and led to the realisation of $151 million in Loss Savings in just 3 weeks.
They also implemented an intelligent Check Kiting fraud prevention for a large tier 1 financial institution by improving capture rate of bad payments, reducing write-offs, reducing payment holds enhancing customer experience. We were able to deliver a clear impact of $15 million in the fraud reduction and were able to reduce their False Positive Rates by 80%
The road ahead for analytics at Scienaptic:
“We have a clientele comprising of 8 clients across various industry verticals. We started with all our clients with small PoCs but over a very short period of time, these have evolved into full-fledged engagements owing to the superlative business impact that we have delivered,” Kulshreshtha said.
The company also has some big future plans. Scienaptic has been working towards strengthening its various use cases within its platform Ether. “(We are working towards building) AI in Credit Underwriting to essentially embed complex AI models for nuanced decisions for loan approvals, credit line management etc., while keeping complete model explainability.We are also looking at embedding AI in Fraud management by using unsupervised fraud models to capture sophisticated fraud and reduce False Positives,” Kulshreshtha said.
“Our goal is to now scale these solutions across all clients. We also have a dedicated Innovations Lab where Data Scientists work towards sharpening AI methodologies to deliver stellar results for clients.”
Scienaptic has received strategic investments from the senior industry leader and former Nasscom Chairman, Pramod Bhasin. The investments have enabled Scienaptic to continue developing its big data platform.
Try deep learning using MATLAB