arrow_back Business analytics on the cloud - a scalable model with R
Banker to the unbanked- story of scale leveraging Data Science, AWS, Scala, Spark
Submitted by Arpit Gupta (@callarpit) on Friday, 16 March 2018
With the online data trail that customer leaves behind. PSPs are leveraging this to understand the ability and intent of these customers to repay back a loan. Who may not even have a bank account and most likely come from tier 2/3 towns in developing countries such as India. This talk is about the ML and data engineering that was put together to provide instant short term credit to millions of consumers(25-50M in India). Later on this product was also launched in 18 other countries such as Brazil, Russia, Columbia, Poland etc adding another 50M consumers. The process involved digital transformation of a PSP by moving from on premises Colo based DC to AWS(Mumbai) and building the best of breed FinTech Architecture that saw two spikes in an year. 1st was the M&A of 2 PSPs resulting in multiple fold increase in traffic(1B + 2.5B TRX/month) and second was the demonetization in India that lead to several (X single digit) times traffic post Nov 8, 2016. An architecture that is able to handle unprecedented scale in a short duration while offering one of its kind checkout credit using digital channels.
It would be good for data and FinTech professionals who are interested in use case of ML and data for credit, lending, risk and related areas. It is also good for people looking to build scale using open source technologies and public cloud services such as Redshift, S3, EMR etc
Talk about credit and lending business
Digital proxy for lending in absence of credit socres, using RFM analysis of customer transactions on payment gateway
Tapping the unbanked population from tier 2/3 towns in India who do not have cards o bank accounts and may use COD, coupons or wallets to make purchases
Leveraging ML and data platforms such as SparkML, Python Pandas, Scala, AWS to build a PaYLater product
Takeaway how to build a data and ML product
Choose the right technology to build the right solution
Solve the lending problem for millions of unbanked peopel in India
Build one of the finest FinTech architectures on a public cloud
Come with an open mind :)
See how FinTech is disrupting the tradition lending market using digitial footprint of consumers with data and ML.
Arpit is a serial entrepreneur and the process started when Arpit worked his way in grade 6 to teach himself programming and write business applications for SMBs in accounting, tax and inventory management. Slowly the business ramped up into training, networking solutions and coaching as he graduated from high school. In mid 2000s, Arpit scaled his outsourcing product consulting to Japan building embedded and network software for automotive and telecom companies. This business grew up into marketing research, HR manpower consulting, corporate training across India and Japan. Both these initiatives had a successful exit(acquired by iTT) and created sustained triple digit growth. Most recently Arpit help found a SAAS company AA that was bootstrapped in the US(SF, TX). Arpit previously lead product engineering teams in Sr leaderships roles at AMZN, YHOO, ORCL, RAX etc. At AMZN he launched SAAS enterprise and consumer products, at YHOO he built trust and safety organization, launched several search and AdTech products for 4 screens. He joined as a founder leader of new communication BU at ORCL scaling it to 250 people organization. RAX invited Arpit to build the analytics sciences organization and KPI for the hybrid cloud.