About Me
Data Science professional specializing in robust engineering infrastructures. Competent in deploying pipeline logic patterns, multi-source Data Warehouse relational blueprints, and high-performance Machine Learning setups explicitly layered for Explainable Artificial Intelligence transparency metrics.
Experience
Data Science & Machine Learning Intern
- Developed quantitative models and evaluated algorithms including LightGBM and Random Forest for market data analysis.
- Extracted and visualized SHAP feature importance metrics to ensure explainability in predictive outcomes.
- Delivered consistent technical project assignments, managed pipeline updates, and maintained a robust engineering portfolio.
Projects Portfolio
Nexvion Copilot
- Architected a real-time visual AI assistant Chrome extension utilizing Manifest V3, Shadow DOM overlays, and offscreen canvas processing.
- Built a high-performance Python backend using FastAPI and WebSocket protocol handlers for seamless, low-latency relay execution.
- Engineered a 'Ghost Cursor' that provides non-intrusive, step-by-step navigational guidance without requiring pre-made static tutorials.
- Deployed an optimized landing page using HTML5 and Tailwind CSS, successfully launching the bootstrapped product on Product Hunt.
Explainable Ensemble Framework for Adaptive Financial Market Prediction
- Engineered a multi-layered ML architecture combining unsupervised market regime clustering with deep supervised ensemble classifiers (LightGBM, Random Forest, XGBoost).
- Built an automated data ingestion framework connecting live market web sockets (Binance, macro indicators) with automated UTC synchronization.
- Deployed a full-stack dashboard featuring SHAP metrics and LIME frameworks to provide real-time Explainable AI (XAI) feature transparency.
E-Commerce DWBI & Operational Analytics System
- Designed a multi-dimensional relational database schema using a customized Star Schema consisting of explicitly defined Fact Tables and Dimensions.
- Developed visual data engineering ETL workflows via KNIME Analytics Platform containing file ingestion, missing-value handlers, and PostgreSQL writing mechanisms.