Puneet Bhasin

Developed strategy and predictive models for a global media conglomerate for customer retention, profile-based pricing and customer behavior. Designed price sensitivity model that predicts customer behavior for upcoming price hike scenario. Implemented automated pricing and retention strategy design to maximize retention and revenue. Designed and implemented Predictive Model to identify potential win back customers. Built a Supplier Risk monitoring solution for a Fortune 500 Pharmaceutical client to manage more than 50,000 suppliers across 7 different risk themes. Improved the match rate of Supplier database by 34% using a custom NLP based match algorithm. Implemented continuous ingestion and delivery (CI/CD) pipeline. Developed a cost-effective retrieval-augmented generation (RAG) system using Langchain and Llama 2. Developed a LLM superimposed graph network framework. Built a Transaction profiling system for a fortune 500 Financial services client.

Developed, Back-tested and deployed trading algorithms to run fully unsupervised and Serverless on cloud thru AWS API Gateway and lambda yielding annual return of 53% from Oct 2021. Achieved 70% reduction in capital requirements by maximizing capital utilization through inter-algorithm interactions.

Supported product improvement and Marketing strategy by delivering actionable recommendations through Aspect based Sentiment analysis using BERT model for customer reviews. Reduced data entry errors by 86% by automating data capture via computer vision-based Image to text extraction model. Created Budgeting and workforce plans to manage workload, expenses and hiring. Optimized task allocation to minimize the distance and travel time through Geo-location mapping using Google Maps API resulting in 22% increase in efficiency. Analyzed profitability and Risk metrics across different product profiles and competition. Designed and Automated periodic reports/dashboards to monitor product, demographic, employee and data quality level KPIs. Managed a team of 4 data analysts.

Developed Acquisition Scorecard model for Analyzing Acquisition Risk for Unsecured Loan applicants resulting in reduction of onboarding time by 74%. Designed a Customer Credibility Analysis model using Fuzzy text matching for customer data captured thru alternate sources. Developed Cross-Sell Propensity segmentation model to target housing loan customers from existing Personal loan portfolio. Implemented Risk based pricing through Customer Risk Profiling resulting in reduction of delinquency by 16% and increase in top-up cross sell volume by 6%. Identified growth opportunities through Geo-Profiling Analysis. Designed and implemented Credit data warehousing ETL Pipeline for Bureau Response data to create 2500+ variables.

Examined trends in data to support business requirements. Performed data extraction and sampling for application testing.