Noopur Divekar

About

Building things that solve problems and drive impact.

I’m a Data Scientist and AI Engineer who recently graduated with a Master’s in Data Science from Indiana University Bloomington (May 2026).

I specialize in architecting end-to-end Data and AI projects that automate manual processes and transform raw information into actionable business intelligence. From engineering ML pipelines on AWS, building multi-agent AI full-stack systems to building Power BI dashboards that mitigate nearly $1B in operational risk, I bring a rare blend of technical depth and business acumen. I’ve worked across startups (Project 990 Inc.), Fortune 500 companies (Zoetis), administration (IU Admissions) and academic research (The Millennium Project).

My approach is centered on active stakeholder collaboration: I deep-dive into the problem, engineer the right solution, and distill technical findings into clear, persuasive narratives that drive decisions. I build solutions that matter.

Experience

Where I’ve made impact.

Led 2 analysts to architect a prompt-agnostic LLM classification pipeline (Llama 3.3 70B, Gemma, Tree-of-Thought reasoning) to categorize 300K+ nonprofit mission statements into 27 NTEE codes, reducing manual classification by ~83%.
Architected a regex-based text extraction pipeline to strip IRS boilerplate and surface clean mission statement data from raw 990 filings, establishing the high-fidelity data foundation required for reliable downstream NTEE classification.
Built Tableau dashboards on millions of IRS 990 & 990PF records to map funder similarity networks and charitable giving patterns, enabling data-driven funding strategy for grant officers and nonprofit researchers.

Surfaced $994M in potential annual operational risk by engineering Power BI dashboards and reporting tool on animal health pharmaceutical supply chain data, equipping 10+ senior leaders to drive strategic capacity utilization decisions.
Eliminated siloed capacity reporting by consolidating 4+ SQL servers, SAP ECC, and Excel sources into a unified Power BI data model with the first standardized capacity planning formula across packaging, and formulation teams.
Drove $345M in cost savings by deploying an interactive Power BI scenario simulation tool with advanced DAX, enabling leaders to model OEE improvements, line shifts, and demand spikes to inform equipment upgrade decisions.

Orchestrated and delivered IU Information presentations to diverse stakeholder groups including over 4000 prospective students, articulating university value propositions to drive enrollment interest.
Served as the primary liaison for large-scale admissions events, coordinating real-time communications between staff and attendees.
Translated complex outreach and recruitment data of last 3 years into actionable business insights using advanced Excel and Power BI, presenting findings to senior leadership.

Engineered Random Forest and XGBoost ensemble models in Python to forecast financial budget allocation for consulting clients, achieving 25% improvement in prediction accuracy and enabling more reliable resource planning.
Architected an automated feature engineering pipeline using SQL and Python to preprocess financial datasets, reducing model training time from 6 hrs to 2 hrs and accelerating ML model iteration across client workflows.
Deployed interactive Tableau dashboards monitoring ML model performance and financial KPIs, surfacing insights that contributed to an 18% increase in client conversion rates.

Uncovered 4+ operational bottlenecks in grant disbursement workflows by querying 1,500+ grant records using SQL and Excel, delivering findings that reduced application processing time by ~18% across program departments.
Designed Tableau dashboards tracking Key Risk Indicators (KRIs) and KPIs, including budget burn rate and grant approval velocity, equipping 5 program directors with real-time visibility to support weekly operational risk reviews.
Facilitated UAT for a grant management system rollout, validating 27+ functional requirements and flagging 3 critical data integrity issues pre-deployment to ensure enterprise risk management compliance.

Projects

What I’ve been working on.

ReadyAid: RAG-Based Offline Mobile First-aid Assistant with LLM Streaming

Full-stack Android first-aid app with local RAG pipeline, real-time LLM streaming via FastAPI, ChromaDB, and Ollama.

KotlinJetpack ComposeFastAPIRAGChromaDBOllama

GitHub Demo

Amazon Marketplace Intelligence Platform

Amazon Marketplace Intelligence Platform

End-to-end cloud analytics pipeline: BigQuery ELT with dbt for 1.4M products across 4 Amazon domains with Power BI reporting.

BigQuerydbtPower BIPythonSQLETL Pipeline

GitHub Power BI

Smart Hospital Multi-Agent AI System

5-agent LangGraph system automating clinical logistics: triage, scheduling, pharmacy, lab, and patient agents using Groq LLM.

LangGraphLLMPostgreSQLPythonAgentic AIAPI Integration

GitHub Demo

Telco Customer Retention Optimizer

Causal uplift modeling with Power BI integration to optimize retention strategies using A/B testing.

PythonScikit-LearnUplift ModelingPower BIA/B TestingCausal Analysis

GitHub Power BI

Cloud-Native Customer Churn Prediction Pipeline

Full AWS pipeline: PySpark ETL, feature engineering, XGBoost on SageMaker Canvas, Power BI dashboard.

AWSPySparkXGBoostSageMakerPower BI

GitHub

Super Bowl Win Probability Model

Logistic regression model predicting Super Bowl outcomes using historical NFL statistics.

PythonScikit-LearnLogistic RegressionSQLnfl_data_py

GitHub

Global Policy Scenario Dashboard

Power BI dashboard with DAX what-if sliders and sensitivity analysis tracking 29 SOFI indicators for a UN think tank.

Power BIDAXPythonSensitivity AnalysisAdvanced ExcelWhat-If Analysis

NDA Protected Power BI Website Credit

AI-Powered YouTube Learning Path Generator

NLP-powered system that structures YouTube videos into personalized learning paths using TF-IDF and cosine similarity.

PythonYouTube APINLTKNetworkXTF-IDF

GitHub

Global Internet Penetration Trend Analysis

World-scale analysis of internet usage data with interactive Plotly visualizations and statistical analysis.

PythonSQLPlotlyK-Means ClusteringSQLite

GitHub

Diet Conversational AI — LLM Fine-Tuning

Fine-tuned Llama 3.2 on nutrition QA datasets using QLoRA and HuggingFace for a dietary advice chatbot.

PyTorchHuggingFaceFlan-T5Weights & BiasesNLP

GitHub

Recommendations

What it’s like to have me on the team.

“

I had the pleasure of working with Noopur on a visualization project of the State of the Future Index (SOFI) of the Millennium Project, as part of her Information Visualization course at Indiana University. The task was to create an interacting model to show the SOFI usefulness to analysis and policy making. As a team leader, Noopur demonstrated exceptional leadership both in communicated with us as a client, as well as guiding the project to a professional-grade deliverable. She proved an exceptional understanding of the large, deeply interconnected dataset and the goal, engineering a sophisticated solution using advanced Power BI and complex DAX. An interactive ‘What-If’ included in the model allows to simulate various scenarios compounding environmental changes in real-time. Her ability to execute such a complex model under tight deadlines proves exceptional combination of technical and leadership skills. I would definitely work with her again and highly recommend her for any data science or analytics role.

Elizabeth Florescu Director of Research, The Millennium Project Project Mentor

“

I highly recommend Noopur for her outstanding skills in data analytics, especially with Power BI. During her internship, Noopur demonstrated impressive technical expertise and initiative by developing a complex capacity utilization dashboard in record time. Her ability to quickly understand business requirements, translate them into actionable insights, and deliver a polished, interactive dashboard exceeded expectations. Noopur combines strong analytical thinking with a hands-on approach to problem-solving. Her proficiency with Power BI — including data modeling, visualization, and DAX — was evident throughout the project, and the dashboard she built has become an invaluable tool for our team. I have no doubt that Noopur will continue to excel and add value wherever she goes. I wholeheartedly endorse her Power BI and data analytics skills!

Steven Rivard Senior Team Leader — KZO NextGen (SAP S/4 Hana), Zoetis Manager at recent internship

Achievements

Milestones & recognition.

🏆

Luddy Hackathon Winner

Second runner-up — designed and deployed the Smart Hospital Multi-Agent AI System (SHMAS) within a strict 48-hour deadline, implementing a 5-agent architecture to automate clinical logistics.

September 2025

📖

Conference Publication — Multicon 2023 International Conference

Published and presented: “Hybrid Approach of Seed Spreading and Weed Plucking for Enhanced Crop Yields using Machine Learning.” Featured in Industrial Internet of Things for Responsible Technology.

ISBN 9781032829401 · May 2023

Building things that solve problems and drive impact.

Where I’ve made impact.

Data Scientist

Data Analytics Intern

Graduate Data Analyst and Assistant

Data Science Intern

Data Science & Business Analytics Intern

What I bring to the table.

What I’ve been working on.

ReadyAid: RAG-Based Offline Mobile First-aid Assistant with LLM Streaming

Amazon Marketplace Intelligence Platform

Smart Hospital Multi-Agent AI System

Telco Customer Retention Optimizer

Cloud-Native Customer Churn Prediction Pipeline

Super Bowl Win Probability Model

Global Policy Scenario Dashboard

AI-Powered YouTube Learning Path Generator

Global Internet Penetration Trend Analysis

Diet Conversational AI — LLM Fine-Tuning

Industry-validated credentials.

Power BI Data Analyst Associate (PL-300)

Fabric Analytics Engineer Associate (DP-600)

Associate Data Analyst in SQL

Data Analyst in Power BI

Data Science Professional Certificate

Academic foundation.

Master of Science, Data Science

Bachelor of Engineering, Electronics

What it’s like to have me on the team.

Milestones & recognition.

Luddy Hackathon Winner

Conference Publication — Multicon 2023 International Conference

Got a role in mind or just want to connect? Let’s talk.