9 Years of experience in Machine Learning & AI across domains of Banking, Fintech E-commerce & Hyper Local delivery, Credit Risk
Hi, I’m
Bhargava Reddy
Lead Data Scientist
Hyderabad
Summary
Overview
9
years of professional experience
Work History
Scienaptic AI
Lead Data Scientist
01.2023 - Current
Job overview
- Have onboarded 30+ Credit Unions by leading a team of 5 data scientists: doubling the live clients in an year, using custom machine learning credit risk models ( Reject Inference, Probability of Default, Adverse Impact Ratios) & standard credit risk strategies using bureau & 3rd party information for Credit Underwriting loan applications in the US.
- Worked with Engagement Leadership to develop quarterly roadmaps based on impact, effort and test coordinations for Credit Unions which lead to 400bps increase in average loan approvals.
- Utilized visualization ( QuickSight ) and analytics tools to analyze and process complex data sets & Identified, measured and recommended improvement strategies for KPIs across different credit products.
- Developed intricate Automated QA process for the deployment of models, complex bureau rules in the credit policies & strategies.
GoKwik
Principal Data Scientist
07.2021 - 07.2022
Job overview
- Lead a team of 4 data scientists & data engineers in the Applied AI team providing Machine Learning Solutions to tackle the RTO & check out solutions for Indian D2C brands.
- Lead the development of a python package called Jarvis that automates feature engineering, feature reduction and tunes the hyper parameters for tree based models. All the clients that required a custom model were served using Jarvis.
- Built Network Ensemble models using BERT, hypothesized by the interventions that need to be taken to stop RTO based on Address, Product Information, Customer Information as a multi faceted solution that is applied to the whole network of GoKwik that reduced RTO by 640bps.
OKCredit
Senior Data Scientist
02.2020 - 06.2021
Job overview
- Transaction addition via voice in Hindi: Trained interactive voice feature that enables users to add transactions via voice in OkCredit app. Was able to achieve 8.2 WER with IIT Madras Hindi ASR Challenge data using Kaldi TDNN + RNNLM.
- Experimented with Quartznet framework (end2end) by transcribing 100 hours of research calls + 100 hours of text to speech data from OKcredit merchants.
- Khata Onboarding via images: Have done a pilot with Roboflow to evaluate the scope of Automatic Khata Onboarding via images. Have built an object detection model using Yolo v5 on 10,000 tagged images obtained from OKCredit app.
- Auto Language Detection Model: Have developed a rule engine based real time Auto Language Detection model in Kubernetes + Seldon.io improving the funnel drop of vernacular languages by 80 bps.
Swiggy
Senior Data Scientist
06.2018 - 02.2020
Job overview
- Customer Fraud Detection Model: Have built an anomaly detection model using Isolation Forest, model was deployed real time to detect fraud and reduced the cost of reimbursement per order by 9 Rs. Had to bring in kafka data streaming into use during deployment to capture fraudsters making multiple requests in a short span of 1-2 minutes post the initial version.
- Driver Fraud Detection Model: Have built a map matching algorithm using HMM to detect fraud by delivery partners making use of GPS anomaly points in route under bridges, overcrowded streets etc... to earn more by tricking the distance tracking system.
- Restaurant Stress Model: Have deployed Restaurant Stress Model using GBM to predict the restaurants that would be under stress during the breakfast/lunch/snacks/dinner slots. Restaurants that have high similarity index are shown to the users in listing instead of the original intended restaurants, upholding the food quality and NPS of the app.
Citibank
Data Scientist
06.2015 - 06.2018
Job overview
- As a part of the business goal to optimize phone channel retention targeting, to move away from a segmentation framework, the following models were spear headed that enabled the Customer Level Modeling Engine: Probability of Retention model, Annual Spend Model, introduced Lend + Spend combination offers, improving average time handled per each call.
Education
Indian Institute of TechnologyKharagpur
Master of Science from Mining Engineering
04.2001
University Overview
- CGPA 9.1/10.0
- One among the two students to score an SGPA of 10/10 in 2/10 semesters and was awarded certificate of merit by the Dean of Undergraduate Studies, IIT Kharagpur.
- Was awarded Institute Silver Medal for being the best student in order of merit among the graduating students.
Skills
Python, SQL, SAS, Spark, Scala
Tensorflow, Keras, OpenCV,scikit-learn, Kubernetes, Dockers
Machine Learning, Deep Learning, NLP, Recommendation Engine, Credit Risk
Classification,ASR, Segmentation, Object Detection, Recommendations, Thought Leadership, Mentoring
Timeline
Lead Data Scientist
Scienaptic AI
01.2023 - Current
Principal Data Scientist
GoKwik
07.2021 - 07.2022
Senior Data Scientist
OKCredit
02.2020 - 06.2021
Senior Data Scientist
Swiggy
06.2018 - 02.2020
Data Scientist
Citibank
06.2015 - 06.2018
Indian Institute of Technology
Master of Science from Mining Engineering
04.2001