Product developed to risk score anomalous and fraudulent transaction happening across Banks Platforms.

A project to determine Fraud and Outlier tranasctions in Unsupervised setting where target labels were not supplied.

The Project was built with the help of detailed Statistical and exploratory analysis of data to understand distribution of transactions.

Extensive Literature Research was done to imitate fraud patterns in Cross Border Payments.

Swift messages of type MT103 were primarily studied.

A Behavioural model of Normal transactions was developed to arrive at threshsholds that could help in risk scoring any anomolous transations.

This was a White Box model using Logistic regression among other Algorithm like Time-series etc.

It is a success and being upscaled to varoius Geographies and Clients.

Worked in a team in the capcity of individual contributor.

I worked as the Lead Data scientist very closely with Business to get the requirements fufilled and expections met.

An inhouse Cahtbot to suffice current requirement of product recommendations and advance feedback from self-learning NLP engine that could learn over time and suggest value added option to the cleint Mangers.

An MPV was developed and exposed as a rest service in Docker this is a futuristic model that has number of variations that are bieng considered including classical NLP based solutions using proximity matrix and multiclass logistic regresion.

I an involded in soltution design,development,testing as well optimisation of model.

This is an HR product that was developed to Guage employees sentiments on varoius employee lifecycle.

This is an extensive NLP pruduct wherein Metadata and Dictionary creation was done.

Number of downstream solutions inculding Attrition,Appraisal analysis/Prediction were developed.

Description Performing descriptive analysis and exploratory data analysis on census Data on HDI to Come up with parameters effecting per-capita income.

Responsibility:- • Building multivariate linear regression model using Ordinary Least Square method. • Using R for model building. • Providing Demos and mock selling of solution to internal leadership.

Working on Credit card dataset to come up with customer loyalty insights. Segmenting the customers on the basis on credit card purchases, finding opportunities for cross sell and up-sell.

Worked on a spam-ham Classifier using Random Forest.

Performing descriptive analysis and exploratory data analysis on census Data on HDI to come up with parameters effecting per-capita income.