About
Nitesh Agarwal is a Senior Data Scientist at Meta (Product Analytics), previously Senior Data Scientist (L6) at Amazon. He has 12 years of experience in data science and analytics across e-commerce, consumer electronics, and marketing — currently leading impactful projects for Meta apps using causal inference, A/B testing, experiment design, and automation.
He also works on AI evaluation — testing AI agents that review experiments — and built an experiment review agent with Claude AI that saves 80% of time in experiment design and reviews. He built SeniorScamBench, a benchmark for how well scam-detection systems protect older adults. Earlier, he spent 5 years in data science consulting for U.S. startups and helped build data science teams. He holds an M.S. in Business Analytics and Data Science from the University of Cincinnati.
What I do
Education
M.S., Business Analytics & Data Science — University of Cincinnati, Carl H. Lindner College of Business, 2019 (4.0 GPA)
B.E., Electrical & Electronics Engineering — Vellore Institute of Technology, 2013 (4.0 GPA)
Certifications
Introduction to Generative AI — Google, Aug 2023. Plus 12 DataCamp certifications in Python, R, and machine learning (2018–2019).
Languages
Experience
Tap a role to expand it.
Data Scientist, Product Analytics
▾
- Led impact analysis of 15 user actions (1B+ users) using fixed-effects regression; thread customization drove +8.9M daily sessions, influencing the H1'26 roadmap and $10M+ in engineering allocation across 14 projects.
- Analysis and recommendations for 8+ A/B tests; launches lifted sessions (+0.04%), sticker sends (+4.7%), and GIF senders (+10.9%) across 500M+ DAU.
- Built experiment automation tooling and dashboards for 8 teams using Claude and internal Meta AI tools; cut experiment review time from 2+ hours to ~20 minutes and established centralized measurement governance.
Data Scientist Manager (L6)
▾
- Created a propensity score model and A/B testing framework, rolling out a digital engine solution for AWS cloud marketing.
- Improved regional customer lifetime value models, increasing accuracy by 50%, optimizing marketing investment and SMB targeting.
- Built reporting tracking 120+ metrics, saving 4.7k man-hours annually and $350K for delivery businesses.
Business Intelligence Engineer Manager
▾
Business Intelligence Engineer-II
▾
Analytics Consultant
▾
- Led fraud analytics for UIDAI, the world's largest biometric ID system — team lead for the fraud analytics division, presenting findings to UIDAI's CEO and leadership team.
- Built an operator scoring framework and machine learning models to identify suspicious operators and enrollment centers, plus reports and rules to flag suspicious activity.
Senior Business Analyst
▾
- U.S. car market analysis; used conjoint analysis to automate new-car pricing based on customers' willingness to pay.
- Segmentation and logistic regression models to identify customers likely to churn and target them with relevant offers.
Decision Scientist
▾
- In-store shopping behavior insights for a major retailer, integrating POS transactions, planograms, and product data; checkout-lane optimization and on-shelf availability alerts.
- Customer segmentation based on internet usage behavior to recommend the most relevant videos.
Hiring a data scientist?
I'm always happy to hear about interesting data science roles.
Get in touch →Research & projects
Solo first-author paper in progress — a performance, quality, and cost framework for testing AI agents that review experiments. Targeting IEEE venues in 2027.
A benchmark and detection app for elder-targeted scams. Phase 1 finding: a standard SMS-spam classifier catches 97.6% of public SMS scams but only 28% of elder-targeted ones. Ask HALT is being built to close that gap.
- Operator Scoring Framework, UIDAI (Government of India) — flagged suspicious enrollment operators; 5,000+ operators blacklisted.
- Authentication Device Behavior, UIDAI — flagged suspicious biometric auth devices; 5,000+ devices blocked.
- Customer Sentiment Analysis, AT&T — social listening and segmentation to lift customer engagement.
- Customer Segmentation & Churn, Godrej Nature's Basket — RFM segmentation and a churn model to retain high-value customers.
- Car Price Optimization, Sonic Automotive — conjoint analysis for willingness-to-pay-based pricing.
- In-Store Behavior Analysis, Walmart — checkout-lane optimization and stock-out alerts; +2% sales.
Research collaboration
Working on experimentation, causal inference, or AI evaluation? Let's talk.
Start a conversation →Review contribution
Invited peer reviewer in AI evaluation, experimentation, and causal inference. Professional Member of IEEE.
Reviewer profiles
OpenReview Google Scholar ORCID Web of Science
Review invitation
I review manuscripts on AI evaluation, experimentation, and causal inference.
Invite me to review →Editorial inquiries
Considering me for an editorial board or guest editor role? I'd love to hear from you.
Reach out →Contact
The best way to reach me is email: hello@datawithnitesh.com. For press and expert-comment requests, see my Qwoted profile.
Hiring a data scientist?
I'm always happy to hear about interesting data science roles.
Get in touch →Review invitation
I review manuscripts on AI evaluation, experimentation, and causal inference.
Invite me to review →Editorial inquiries
Considering me for an editorial board or guest editor role? I'd love to hear from you.
Reach out →Research collaboration
Working on experimentation, causal inference, or AI evaluation? Let's talk.
Start a conversation →