┌──────────────────────────────────────────────────────────────────┐
│ > SELECT * FROM analysts WHERE curious = TRUE │
│ AND detail_oriented = TRUE AND results_driven = TRUE; │
│ │
│ 1 row returned. ✓ │
└──────────────────────────────────────────────────────────────────┘
Data analyst who treats every dataset like a case to close — not just a chart to make. I've reverse-engineered why YouTube videos go viral, mapped India's data-analyst hiring market, and rebuilt retail dashboards that surface risk before it hits the P&L.
Currently seeking Data Analyst roles where SQL, Python, and Power BI turn ambiguity into a decision someone can act on Monday morning.
| 📊 Datasets Analyzed | 🎯 Dashboards Shipped | 📈 Records Processed | 🏙️ Markets Covered |
|---|---|---|---|
| 5 end-to-end | 4 interactive | 600K+ rows | India + Global |
Core stack — job-ready
Pandas ▓▓▓▓▓▓▓▓▓▓ Job-Ready
Data Cleaning ▓▓▓▓▓▓▓▓▓▓ Strong
NumPy ▓▓▓▓▓▓▓▓▓▓ Intermediate
Python ▓▓▓▓▓▓▓▓▓░ Lower Intermediate
Matplotlib ▓▓▓▓▓▓▓▓▓░ Intermediate
Excel ▓▓▓▓▓▓▓▓▓░ Advanced
Power BI ▓▓▓▓▓▓▓▓▓░ Intermediate–Advanced
Data Visualization ▓▓▓▓▓▓▓▓▓░ Strong
SQL ▓▓▓▓▓▓▓▓▓░ Intermediate
Git / GitHub ▓▓▓▓▓▓▓▓░░ Intermediate
Communication ▓▓▓▓▓▓▓▓░░ Good
{
"querying": ["SQL — Joins, CTEs, Window Functions, Subqueries, Self-Joins"],
"spreadsheets": ["Excel — Pivot Tables, Power Query, Dynamic Dashboards"],
"visualization": ["Power BI — Data Modeling, DAX, Interactive Reports", "Matplotlib"],
"programming": ["Python — Pandas, NumPy, Matplotlib"],
"applied_ai": ["LLM-assisted classification, zero-shot labeling pipelines"],
"process": ["Data Cleaning", "KPI Design", "Trend Analysis", "ABC Segmentation"]
}
Reverse-engineered what makes long-form content go viral — analyzed 545+ videos (758M+ views) using the YouTube Data API, LLM-based zero-shot classification, and Power BI.
| What I Did | Key Outcome |
|---|---|
| Built end-to-end Python pipeline pulling 545 videos via YouTube Data API v3 | Structured dataset, 9 columns: views, likes, duration, metadata |
| Zero-shot classified every title across 8 strategic dimensions with an LLM | Enriched to 17 columns — zero manual labeling |
| Mapped hook structures, emotional triggers & title patterns | 91.6% of top performers use a "Curiosity Gap" hook |
| Built multi-page Power BI dashboard with a virality recommendation engine | Formula identified: AI/Finance topic + 15–18 word title + Curiosity Gap |
SQL + Power BI analysis of 500+ data analyst job postings across 10+ Indian cities — skills, salaries & hiring trends.
| What I Did | Key Outcome |
|---|---|
| Designed a normalized schema (jobs, skills, jobskills junction table) | Enabled multi-dimensional skill × salary analysis |
| Used self-joins to detect skill co-occurrence | SQL + Python is the top combo (~338 mentions) |
| Salary distribution analysis by city & skill tier | Python/Power BI roles pay 20–30% more than Excel-only |
Skill-gap analysis using NOT IN subqueries |
Mapped an exact upskilling path for entry-level candidates |
End-to-end BI solution analyzing 50,000+ retail transactions across 4 years (2015–2018).
| What I Did | Key Outcome |
|---|---|
MoM & YoY growth via LAG window functions + CTEs |
Revenue grew 50%: $4.8M (2015) → $7.2M (2018) |
| ABC customer segmentation with cumulative window functions | Isolated high-LTV clusters for targeted marketing |
| 3-page interactive Power BI dashboard with DAX measures | West region drives 31% of total revenue |
| Product concentration risk analysis | Top 5 products = outsized revenue share — diversification flagged |
Multi-table SQL extraction + interactive Excel Executive Dashboard with KPIs, slicers, and 7+ chart types.
| What I Did | Key Outcome |
|---|---|
| Multi-table JOIN across 9 tables (sales + production schemas) | Single flat dataset powering the whole dashboard |
| Interactive slicers for Year, State, and Store | Baldwin Bikes drives 68% of $8.58M total revenue |
| Bing map chart showing revenue by state | 2017 peaked at $3.84M — 42% YoY increase |
| Sales rep performance ranking | Top rep contributed $2.93M individually |
Exploratory data analysis on 9,827 Netflix movies spanning 1902–2024 using Python, Pandas & Seaborn.
| What I Did | Key Outcome |
|---|---|
| Full EDA pipeline — loading, cleaning, feature analysis | Drama is most frequent; Action gets the most votes |
| Popularity vs. vote-count correlation analysis | Vote count predicts popularity better than rating |
| Genre and language diversity analysis | English dominates at 77% despite 43 languages present |
| Release-year trend analysis | 2020 was the peak year for releases |
▓▓▓▓▓▓▓░░░ Statistics & A/B Testing Fundamentals
▓▓▓▓▓▓▓░░░ EDA — moving from checklist to intuition
░░░░░░░░░░ Seaborn (queued next)
░░░░░░░░░░ Workflow Automation (queued next)
"Without data, you're just another person with an opinion." — W. Edwards Deming