Lead AI/ML Research Engineer

github

Chanabasgouda Patil  • Peerlist

Worked on customer segmentation for one of the Indian travel application AbhiBus.com which provides transport facility. Worked under Passenger segmentation, Operator Segmentation , RFM Analysis.

AbhiBus - Customer Segmentation

- Business Understanding: Collaborated with stakeholders to understand business operations and align the AI model with the bank's risk management strategies. Conducted exploratory data analysis (EDA) to validate insights and ensure they were consistent with business objectives. - Data Integration and Preprocessing: Conducted data profiling, distribution checks, and handled missing data and outliers to ensure high-quality input for the model. Integrated data from multiple sources, including credit bureaus and financial statements. - Feature Engineering: Selected relevant features through methods like R2 value from DecisionTreeRegressor, optimizing the model's predictive power. - Model Development: Designed and implemented a machine learning model to predict the probability of default and other risk factors using Logistic Regression, Random Forest, and XGBoost. - Model Validation: Performed model validation using techniques such as K-Fold Cross Validation and back-testing to ensure robustness and accuracy. -  Deployment and Integration: Successfully deployed the model into the bank’s existing loan origination system, enabling real-time decision-making in the credit underwriting process. Impact: ------------------------------------------------ - The model reduced loan default rates by accurately identifying high-risk applicants, leading to a 6% increase in overall profits.Improved processing efficiency, reducing underwriting time, and enhancing customer satisfaction. - Enhanced the bank's ability to manage credit risk, resulting in more sustainable lending practices

AI enabled Credit Underwriting system

Responsible for developing and deploying an AI feature on the Havells One application, which is capable of identifying the genuine invoice of the Havells product to prevent fraudulent transactions in the loyalty program and customer service request flow.

AI Invoice Classifier

Worked on Restaurant related dataset.Performed aspect based sentiment analysis to gain the insight from the feedback from the customer w.r.t aspects like service, ambience, food quality etc. parameter.

Aspect Based Sentiment Analysis(ABSA)

Developed AI platform to monitor the assets(AC, Refrigerator, Washing Machine etc) in real time and provided AI services including Predictive Maintenance, Preventive Maintenance, Health Report, Self Diagnosis and monitor customer devices.

Asset Monitoring Platform

This project is in Atlanta Telecommunication (ATNI-Guyana) of US, where proposed and develped solution predicted churning of customer. From the prediction, organisation took a measure to improve the customer service and retain them before switching to other competitors network.

ATNI Guyana - Telecom Customer churn predicti

Worked on Credit & Market risk related project for Bank of America in APAC, EMEA countries.

Bank of America(BoFA)

Developed a product which classifies the incoming tickets. Based on prediction, ticket will be assigned to particular team who going to resolve the ticket.

Intelligent Service Desk(I.S.D) - ITM

I worked on a complex price prediction project for JBS Pilgrim's meat manufacturing company in USA.

JBS Pilgrims - Product Price Prediction

Developed an AI Engine which recommends the best AC settings and automatically changes in AC, leading to better comfort, increased user experience and energy-efficient cooling.

Lloyd AC AI Auto Mode

Developed Chatbot using Dialogflow, Kore.ai to handle HR related conversation which will be integrated with SAP Fiori ,HRMS tools and ticketing tools like ITSM and BMC Remedy.

MOURI-HR Chatbot development

Developed a robust and reliable AI solution which detects errors in the Washing Machine and provides DIY instructions to Lloyd Washing Machine users in the Havells One application, leading to increased user experience.

Self Diagnosis of Error

Worked on developing classification model which predict the ticket category based on text present in ticket description of Tickets.