Data Analyst with hands-on experience in Python, SQL, and BI tools, plus applied Generative AI (RAG, LangChain, LLM integration). Built and shipped 10 end-to-end data projects using self-collected, real-world data — incl…
Data Analyst with hands-on experience in Python, SQL, and BI tools, plus applied Generative AI (RAG, LangChain, LLM integration). Built and shipped 10 end-to-end data projects using self-collected, real-world data — including a retrieval-augmented Q&A system and a live deployed AI analytics tool — rather than pre-cleaned datasets. Skilled in data cleaning, exploratory analysis, dashboard development, and turning messy real-world data into decision-ready insights.
Experience
L
Data Scientist Intern
Luminar Technolab·Ernakulam , Kerala , India
Sep 2025 - Present · 1 year 2 months
<p>Built and deployed machine learning models (regression and classification), improving prediction accuracy by approximately 15–20%.</p><p> Cleaned and processed datasets using Python (Pandas, NumPy), improving data quality and analysis efficiency.</p>
Kochi Restaurant Market: Cuisine Demand-Gap AnalysisProblem<p>If you were deciding which type of restaurant to open in Kochi, which cuisine gives you the best shot — not by gut feel or "everyone opens a Kerala restaurant," but by an actual demand-versus-supply number?</p>Data source & collection method<p>90 restaurant listings scraped from Zomato's structured (JSON-LD) data across 10 real Kochi localities (MG Road, Edappally, Kakkanad, Panampilly Nagar, Kaloor, Palarivattom, Fort Kochi, Kadavanthra, Marine Drive, Ernakulam City) — not a pre-existing dataset. Collected via a Python scraper respecting robots.txt and rate-limited requests.</p>A data-quality finding worth stating upfront<p>63% of the initially scraped listings (57 of 90) were duplicates — the same restaurants appearing under multiple "nearby" localities. This revealed that Zomato's per-area listing widget draws from a shared citywide pool rather than fully distinct per-locality rankings, rather than being a scraping error. After deduplication, this analysis covers Kochi's 33 most prominent, most-reviewed restaurants — a meaningful but not exhaustive slice of the market, and the findings below are scoped accordingly.</p>Approach<ul><li>Python (requests, BeautifulSoup) to scrape and parse structured listing data</li><li>Pandas for cleaning, deduplication, and cuisine-tag parsing</li><li>SQLite for storage and aggregation</li><li>A custom demand-intensity index: rating × log(review_count + 1), which prevents one high-review outlier from dominating an average while still rewarding restaurants that are both popular and well-rated</li><li>A percentile-based gap score per cuisine: demand_percentile − supply_percentile, so cuisines are ranked by demand relative to how many restaurants already compete in that category, not raw counts</li></ul>Key findings<ul><li>Burger (+73.1) and Coffee (+61.5) cuisines show the strongest demand-to-supply gap— the clearest whitespace among Kochi's top-tier restaurants</li><li>Kerala (-73.1) and South Indian (-57.7) are the most oversaturated — despite being the two most common cuisines in the dataset, they rank lowest on demand relative to supply</li><li>Review count and rating are essentially uncorrelated (Pearson r = -0.039) — a restaurant with thousands of reviews is no more likely to be highly rated than one with a few hundred</li><li>63% of scraped listings were duplicates across localities — a finding about platform behavior as much as a data-cleaning step</li></ul>Business recommendation<p>A new Burger or Coffee-focused concept has real, measurable whitespace among Kochi's most prominent restaurants. A new Kerala or South Indian restaurant — the default choice for most new entrants — would be competing in the most crowded, most commoditized segment of the market.</p>Limitations<p>Sample reflects Zomato's "most prominent" listings per locality, not a full census of every restaurant in Kochi. Findings describe patterns among high-visibility restaurants, not the entire market.</p>Tools<p>Python, Pandas, SQLite, Chart.js</p><p></p><p></p>
Certifications
data science and AI
luminar technolab
· Ernakulam , Kerala , India
· 2026
Skills
PythonPandasNumPyMatplotlibSeabornSQLSQLiteMySQLPostgreSQLPower BIDAXData ModelingTableauExcelPivot TablesVLOOKUPMacrosChart.jsRAGRetrieval-Augmented GenerationLangChainGoogle Gemini APIChromaDBPrompt EngineeringLLM-based Data ClassificationScikit-learnRegressionClassificationClusteringExploratory Data AnalysisGitGitHub CLIREST APIsWeb ScrapingBeautifulSoupStreamlitGitHub PagesyfinanceMachine Learning
