Projects

Flip a card to read the summary and open the repo.

Optimized Climate Data Integration with Real Time LLM Querying

SQLite MCP Docker

Overview

Reprocessed planet-scale climate data (ERA5, EDGAR, NOAA, Carnegie) with rigorous wrangling, normalization, and feature engineering—scaling across SQLite + HPC. Built custom MCP servers wired to Anthropic LLMs for context-aware REST querying via prompt-engineered workflows. Shipped a Dockerized MLOps pipeline enabling modular data ops and secure, low-latency SQL for real-time climate intelligence.

View Repo

Telecom Customer Churn Prediction

PySpark Databricks MongoDB

Overview

Built a Docker-backed MongoDB → Databricks/PySpark pipeline and unified usage/billing data with robust feature engineering. Trained and tuned Gradient Boosting (≈80% accuracy) to rank at-risk segments, surfacing churn drivers and prioritizing targeted retention actions.

View Repo

YouTube Social Sentiment Pipeline

C++ ETL YouTube API Python Sentiment (VADER) Power BI

Overview

End-to-end pipeline that ingests YouTube comments with a C++ CLI (libcurl + nlohmann/json + spdlog), writes NDJSON, transforms & scores sentiment in Python (pandas + VADER), and visualizes KPIs in Power BI. Collected 2,222 comments over 150 days; ~42% positive, 30% negative, 28% neutral. Includes batch runner run_verizon.sh and curated CSV/Parquet outputs.

View Repo

Pandemic Impact on Healthcare & Happiness

ML Ensemble learning JHU COVID Data World Happiness Index

Overview

Combined World Happiness, JHU COVID, and vaccination datasets; cleaned, normalized, and engineered features to model well-being vs. health outcomes. Benchmarked Random Forest/Decision Tree/Naive Bayes models—ensembles performed best (~92.5% accuracy) and revealed key non-linear relationships.

View Repo

Target Retail Market Analysis

Qualtrics Tableau Dashboards

Overview

Combined a Qualtrics survey (n=101) with demand signals and delivered Tableau dashboards highlighting price perception, UX friction, and assortment gaps. Shipped clear recommendations across pricing, owned brands, and omnichannel UX with testable actions for measurable lift.

View Repo