Himanshu Saini | Machine Learning Engineer

Himanshu Saini

Machine Learning Engineer

Based in New Delhi, India

About

I am a Data Scientist working at DMI Finance, where I build production-grade Generative AI and machine learning systems for digital lending. My work includes developing LLM-powered solutions such as natural language–to–SQL bots, OCR-based financial document intelligence, and AI-driven chatbots embedded in mobile loan journeys, along with predictive models for affluence estimation and personalization. Previously at RBL Bank, I developed credit risk and income estimation models using Python, PySpark, Databricks, and XGBoost, delivering measurable improvements in risk tagging accuracy and customer profiling within a regulated banking environment. I focus on building scalable, real-world ML and GenAI systems across lending, risk, and customer experience.

Skills

Projects

Credit Portfolio Stress Testing Framework

Developed a portfolio-level credit risk stress-testing system using public loan and macroeconomic data to evaluate how economic shocks amplify losses across credit grades, with extensions to both US and India macro environments.

Decision-Oriented Fraud Risk Ranking under Weak Labels

Designed a fraud risk ranking system under limited review capacity, leveraging Isolation Forest and LOF disagreement with rank-based ensembling, evaluated via Precision@K and compared to a supervised XGBoost upper bound.

ML - Salary Prediction and SHAP values

The project aims at developing machine learning model which can predict salary of the developers all around the world. In this project i have done feature engineering and Exploratory data analysis to get useful features from the data to train the model. At the end SHAP value are also introduced

Stack Overflow Developer Survey 2023 India Region EDA

Exploratory Data Analysis for India region, about the diverse trends in the Indian technological industries .Exploring the responses from developers in India, we see what are the most popular technologies and programming languages and what is the declared level of yearly compensations and more.