data scientist + analytics engineer
β building analytics and ML systems that turn messy data into usable decisions
β strongest zones: healthcare analytics, customer intelligence, NLP, dashboards
β mentored 1800+ working professionals globally
β current focus: applied AI systems with clean pipelines, useful metrics, and decision-ready UX
| π data layer | EDA Β· feature engineering Β· data cleaning Β· SQL Β· data quality |
| π€ modeling layer | classification Β· clustering Β· forecasting Β· scoring Β· evaluation |
| π¬ intelligence layer | NLP Β· embeddings Β· semantic similarity Β· image captioning Β· retrieval |
| π decision layer | dashboards Β· work queues Β· KPI systems Β· storytelling β action |
| ποΈ engineering layer | SQLite Β· Python packages Β· tests Β· CI/CD Β· deployment workflows |
| βοΈ execution | GitHub Β· Streamlit Β· GitHub Pages Β· reproducible documentation |
systems built across analytics, machine learning, NLP, and applied AI
designed for real impact, not just demonstration
| project | what it solves | stack / proof |
|---|---|---|
| Healthcare Claims Intelligence Β· live | PMPM, readmissions, HCC/RAF-style risk, FWA review, high-cost member scoring | Python Β· pandas Β· scikit-learn Β· SQLite Β· Plotly Β· pytest Β· GitHub Pages |
| Claims Denials Revenue Cycle Analytics Β· live | Denial prevention, appeal prioritization, payer friction, underpayment discovery | Python Β· ML scoring Β· SQLite marts Β· data dictionary Β· model cards Β· CI/CD |
| Provider Network Value-Based Care Analytics Β· live | Provider benchmarking, ACO performance, quality scoring, contracting strategy | Python Β· scoring logic Β· custom CSV ingestion Β· tests Β· GitHub Pages |
| Customer Intelligence Platform Β· live | Segmentation, churn prediction, revenue forecasting, retention action planning | Streamlit Β· scikit-learn Β· KMeans Β· churn scoring Β· CSV upload |
| Quora Duplicate Question Detector | Detects semantic duplicate questions with engineered NLP features and embeddings | Sentence Transformers Β· XGBoost Β· 777 features Β· F1 0.8041 Β· ROC-AUC 0.9269 |
| AI Caption Studio Β· live | Captions images, videos, and live camera frames with optional narration | BLIP Β· PyTorch Β· Hugging Face Β· OpenCV Β· WebRTC Β· gTTS |
1. understand the real problem
2. identify signal, ignore noise
3. define metrics that matter
4. build simple before complex
5. apply ML only when useful
6. deliver something usable
β deployed analytics apps across healthcare, customer intelligence, NLP, and productivity
β built projects with data pipelines, scoring logic, dashboards, tests, docs, and deployment
β documenting DSA patterns through a daily LeetCode journal
β learning focus: applied AI, ML systems, analytics engineering, and production-ready storytelling
β solving DSA problems with focus on patterns
β breaking problems before coding
β optimizing for clarity and efficiency
β treating every solution as interview revision material
Good analysis explains. Great analysis drives decisions.


