Data Analyst · SQL, Python, Power BI · Pune, India
I turn business data into decisions people act on. 2.5+ years across D2C, retail and financial services. I built the RFM segmentation and retention models behind ₹50 Cr (~$6M) of client revenue, architected a pipeline running from a custom Shopify app through to leadership dashboards, and audited 30,000+ loan accounts in SQL and Power BI for delinquency and default risk.
Five end-to-end projects on real public datasets. No Titanic, no Iris. Each one clones and runs, each has CI, and each README writes up the bug I hit and how I proved the fix.
| Project | Stack | What it found |
|---|---|---|
| NYC Taxi ETL Pipeline | Python, DuckDB, SQL | ETL over 9.55M real trips with 13 declared data-quality rules. 369,850 bad rows go to a quarantine table tagged with the rule that caught them, so loaded + quarantined = rows_read still reconciles against the source files. |
| IMDb SQL Portfolio | Pure SQL, DuckDB | 15 queries over 141M rows. Kevin Bacon numbers via WITH RECURSIVE: 18,120 actors traversed in 2.2 seconds, no procedural code. The sophomore slump turns out to be real, at -0.208 across 3,664 directors. |
| Superstore Cloud Analytics | AWS S3, Glue, Athena, SQL | Discounts past 20% cost $135,364 and bought no extra volume at all: units per line stay flat at 3.74-3.97 in every band. Capping at 20% lifts profit 47%. |
| Netflix ELT & Data Cleaning | SQL, DuckDB, dbt-style layers | A naive COUNT(DISTINCT country) says 681 countries. There are 117. Staging to intermediate to marts, with 10 assertions holding each cleaning rule in place. |
| Yelp Reviews to Snowflake | S3, Snowflake, Python, SQL | Unhappy customers write 37% more. Exclamation marks turn out to be U-shaped across the rating scale, so they track intensity rather than sentiment. |
Bank Loan Analytics · Excel, Power BI, Tableau, SQL Loan applications, funding, repayments and borrower demographics across the full BI stack.
Data Analyst, HyperScale Leaders Tribe · D2C growth consultancy, 300+ active founders
- Built the RFM segmentation and retention models that directed targeting behind ₹50 Cr (~$6M) in client revenue
- Architected the pipeline running custom Shopify app to Shopify Admin API to Supabase to dashboards, with daily metrics delivered automatically
- Delivered the central reporting layer of an internal ERP, cutting 60+ hours a month of manual reporting
- Owned validation before data entered the ERP: tracing quality issues back to the source system, reconciling spend-to-revenue and ROAS
- Automated reporting workflows in n8n and Pabbly, saving another 40+ hours a month
Data Analyst & Data Scientist (Intern), AI Variant
- Audited 30,000+ loan accounts in SQL and Power BI for delinquency and default-risk patterns
- Built a hyperparameter-tuned SVM bankruptcy-detection model at 93% accuracy, cutting estimated default exposure by around 15%
Football Data Analyst (Freelance) · 925+ statistical match analyses for outlets including Bleacher Report and 433. Grew an analytics community to 17K+ followers and 2.4M+ views.
Data & BI
Engineering
Automation
Techniques · RFM segmentation, cohort and retention analysis, churn modelling, credit risk and delinquency, ETL/ELT pipelines, data quality testing, source-to-target mapping, star-schema modelling
- M.Sc. Data Science, Chandigarh University (2026-2028, ongoing)
- B.Sc. Physics, Savitribai Phule Pune University, 78.78%
- IBM | NASSCOM Data Analyst Professional Certification · HackerRank SQL (Basic / Intermediate / Advanced)
- Going deeper on Snowflake and dbt. The Yelp and Netflix projects above are where that work lands.
If I don't know the answer, I know how to find one.