Default profile banner
VP

Vishal Pateshwari

@vishalpateshwari

Data Science Analyst / Machine Learning Engineer

Gurugram, India

AMERICAN EXPRESSINDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR

Vishal Pateshwari is an experienced Data Science Analyst and Machine Learning Engineer with expertise in developing predictive models. His background includes creating credit risk and payment risk indices using advanced techniques like XGBoost and Logistic Regression. He is proficient in deep learning, computer vision, and NLP, utilizing frameworks such as TensorFlow and PySpark to solve complex data challenges.

Experience

DATA SCIENCE ANALYST

AMERICAN EXPRESS

•Mar 2020 - Present•Gurugram, India

Developed XGBoost model with bureau and amex internal variables to evaluate Credit Risk(default) on New Accounts with an accuracy of 94.8%. Developed XGBoost model which assesses the probability that a merchant’s Credit stress in next 12 months, with an estimated PTI of $4.9Million. Created an Index based on Logistic Regression to assess the credit risk index of the B2B wireless forex payments, estimated to save $12.1Million. Developed Logistic Regression PD Model for two major Market Portfolio.

MACHINE LEARNING ENGINEER

QUANTIPHI ANALYTICS

•May 2019 - Feb 2020•Mumbai, India

Worked for an Insurance Giant to create Customer Segmentation using insurance, lifestyle and health features and scaled up the solution using Gaussian Mixture Model. Developed Basic Recommendor System for segmented customers using Markov Chain.

DEEP LEARNING SUMMER INTERNSHIP

CLIK.AI(PGAM ANALYTICS PVT. LTD.)

•May 2018 - Jul 2018•Gurugram, India

Used Image Processing techniques for Image Skew Correction and devised an algorithm to detect Header and Footer with 83.22% accuracy to clean Image Data. Developed Novel OCR system (based on LeNet-5 Architecture) from Scratch, which converted Images to editable document with 72.3% accuracy on word matching.

MACHINE LEARNING SUMMER INTERNSHIP

INTEGRAND ANALYTICS

•May 2017 - Jul 2017•Gurugram, India

Developed Decision Tree Model for Underwriting efficiency by categorizing data from Readable Financials with an accuracy of 92%, using RegEx and NLP Techniques.

Education

INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR

INTEGRATED MASTER'S

Jan 2014 - Jan 2019•Grade: Cum. GPA: 7.12 / 10

Skills

Machine Learning
Deep Learning
NLP
Computer Vision
Data Structures & Algorithms
Python
SAS
SQL
HiveQL
NumPy
pandas
scikit-learn
Matplotlib
TensorFlow
Keras
PySpark
Hive