Senior Data Scientist | 5+ Years Experience | AI, Advanced Analytics, Machine Learning

Data Scientist with 5 years of client-facing experience and a Master’s in Mathematics from IIT Madras, specializing in data science, ML, AI, and GenAI solutions across Banking & Financial Services, Manufacturing, Insurance, Hospitality, and Retail.

Skilled in building predictive models, optimization engines, forecasting tools, and RAG applications that drive revenue growth, cost reduction, and operational efficiency. Experienced in translating complex business problems into scalable analytics products using Python, SQL, XGBoost, CatBoost, LangChain, Snowflake, DBT, and Tableau.

WORK EXPERIENCE

Data Scientist, QuantumBlack AI, McKinsey & Company, Gurugram, HR Nov 2025 – Ongoing

Credit Risk Model for Personal Loans (CASA and Bureau)

• Developed two Gradient Boosting-based credit risk models for personal loan underwriting, a Bureau model using 2,000+ features engineered from raw bureau PDF data (Gini: 0.48) and a CASA model using 1,000+ features derived from transaction and account balance data (Gini: 0.45), to capture complementary risk signals across bureau and behavioral data.
• Validated both models via crosstab analysis on modelling and out-of-time data to calibrate cutoff decisions, increasing approval rate from 11% to 22.5% while holding the bad rate below the business-mandated 3.1% threshold.
• Layered obligation filters and loan-sizing logic on top of model scores to right-size loan amounts per customer risk profile, scaling monthly disbursements from 800M to 1.3B LKR.

Yield and Throughput Optimization for Chemical Producer

• Designed a throughput optimization solution for a chemical producer by developing a differential optimizer built on top of an XGBoost regression model, increasing plant throughput by 8 TPD and driving ₹2.3 Cr in annualized revenue impact.
• Led the development of sales forecasting model using Light GBM for runner products to support production planning, enabling optimized planning decisions and contributing to a 1.3% reduction in variable costs.
• Created a proof-of-concept Root Cause Analyzer chatbot to analyze DCS data and surface potential drivers of reduced productivity, accelerating issue diagnosis for operations teams.
• Applied SHAP-based interpretability and sensitivity analysis to translate nonlinear XGBoost recommendations into operational levers for plant teams.

Data Scientist, Sixty Four Squares, Pune, MH May 2022 – Sep 2025

Events Attendance Forecast

• Engineered an ensemble of XGBoost and Random Forest models to predict the attendance of future events, resulting in a 30% increase in waitlist-to-booking conversions.
• Automated complex SQL scripts in Snowflake via DBT to streamline data processing and feature engineering for 12 terabytes of data, driving a 2% increase in event attendance by enabling more accurate forecasting.
• Deployed an executive-facing dashboard in Tableau to visualize a two-week attendance forecast, improving staffing decisions and a 3.2% increase in revenue prediction accuracy.

Application Scoring for Membership Approvals

• Improved high-value member identification by 40% by developing CatBoost models in Python to predict first-year visits, spend, and retention for membership applicants.
• Enabled a 25% rise in high-value member approvals by engineering a scoring pipeline with Pandas and Scikit-learn and integrating it into the membership review workflow.

RAG-Based Document Processing Application

• Collaborated on a Retrieval-Augmented Generation (RAG) application in Python to automate insight extraction from unstructured interview transcripts, enabling a 40% reduction in manual review time and accelerated qualitative research workflows.
• Achieved 30% faster semantic retrieval and improved output accuracy by engineering an ingestion-to-response pipeline employing LangChain, Streamlit, OpenAI API, and FAISS for vector search and context-aware question answering.

Marketing Campaign Optimization via Logistic Regression Ensemble

• Designed and implemented a logistic regression ensemble in SAS to optimize product targeting, yielding a 3% increase in conversion rates across a prospect base of over 8 million users.
• Analyzed and delivered a fully scored prospect dataset to the marketing team to enable targeted outreach, helping achieve a 20% reduction in campaign planning time and more precise audience segmentation aligned with business goals.

Research Intern, Indian Institute of Technology Madras, Chennai, TN Jan 2022 - May 2022

Advanced Bayesian Estimation for Linear Models

• Performed Bayesian inference using PyStan to implement the Hamiltonian Monte Carlo (HMC) algorithm for linear regression modeling, resulting in a 35% improvement in parameter estimation accuracy compared to OLS benchmarks.