00Health Analytics · Model Governance
NHANES Multi-Source Health Analytics
Modern reconstruction of a recovered CSIT-558 Data Mining project using CDC/NHANES public-use data, with reproducible cohort construction, interpretable logistic regression, calibration, subgroup validation, and governed challenger-model testing.
- Python
- Pandas
- scikit-learn
- Logistic Regression
- Calibration
- Model Governance
Preferred benchmark ROC-AUC 0.668 · nonlinear challengers rejected after subgroup guardrail failures
01Document Intelligence · NLP · Power BI
Executive Orders Policy Intelligence & Analytics
Provenance-aware document-intelligence project for public U.S. presidential documents, combining deterministic priority rules with SentenceTransformer semantic similarity, a 14-domain taxonomy, alternative-domain signals, review/source-quality heuristics, and Power BI-oriented analytics.
- Python
- SentenceTransformers
- all-MiniLM-L6-v2
- BeautifulSoup
- pypdf
- Power BI
- DAX
Recovered V1–V3 source lineage · public sanitized implementation · primary/secondary/tertiary domain signals · explicit provenance boundaries · no unsupported accuracy claims
01AGenerative AI · NLP
GenAI Amazon Video Game Reviews
Reconstructed and enhanced Amazon Reviews 2023 NLP pipeline using FLAN-T5 for prompt-conditioned cleaning, one-sentence summarization, and feature/issue extraction, with BERT-based 1–5 star sentiment inference and explicit ordinal evaluation.
- Python
- Hugging Face
- FLAN-T5
- BERT
- scikit-learn
- pytest
34-test suite across Python 3.11–3.13 · token-aware truncation · CSV/JSON/JSONL outputs · Ruff + dependency audit + full runtime compatibility gate
01BDeep Learning · Computer Vision
Cats vs Dogs CNN
Transparent TensorFlow/Keras reconstruction of a recovered 20,000-image cat-vs-dog experiment, with memory-efficient tf.data ingestion, explicit class mapping, image-integrity checks, structured validation metrics, and evidence-led overfitting analysis.
- TensorFlow
- Keras
- CNN
- Computer Vision
- scikit-learn
- pytest
Historical: 97.23% train · 82.30% validation · 25-test current suite · precision/recall/F1/ROC-AUC · Ruff + dependency audit + TensorFlow runtime CI
02Behavioral Analytics · Unsupervised ML
Student Routines & Mood Clustering
Reconstructed behavioral-analytics pipeline that converts semi-structured Excel activity diaries into auditable event-level and participant-day data, then clusters behavioral routine features with HDBSCAN while keeping mood outside the clustering feature space. It also supports TF-IDF, optional Sentence-BERT weekday/weekend comparison, deterministic t-SNE visualization, data-quality diagnostics, and descriptive mood-association analysis.
- Python
- Pandas
- HDBSCAN
- Sentence-BERT
- TF-IDF
- t-SNE
- SciPy
33-test suite across Python 3.11–3.13 · end-to-end workbook/output validation · HDBSCAN diagnostics · Ruff + dependency audit · authentic source workbook not published
03Operational Automation · ITS/SCATS
NJDOT Green-Time Anomaly Detection
Sanitized Python operator-support workflow for deterministic Green-Time screening against row-level baselines, configurable severity thresholds, operational input-quality review, daily batch processing, and auditable Excel outputs.
- Python
- Pandas
- openpyxl
- ITS/SCATS
- Operational Analytics
- Data Quality
23-test suite across Python 3.11–3.13 · reason-coded classifications · Ruff + dependency audit · no live operational data
04FinTech Application
Wealth Compass
Browser-based personal-finance analytics application with CSV and searchable-PDF ingestion, transaction normalization, monthly cash-flow analysis, robust spending baselines, deviation screening, goal and investment scenario modeling, and evidence-calibrated recommendation signals.
- TypeScript
- Vite
- PDF.js
- Financial Analytics
- Robust Statistics
- Vitest
Browser-side file processing · deterministic npm lockfile · tests + production build + dependency audit in Node.js 24 CI · searchable-text PDF only, no OCR
05Backend Engineering · PostgreSQL
Course Registration REST API
Reconstructed and substantially enhanced Flask/SQLAlchemy registration backend with terms, sections, prerequisites, schedule validation, FIFO waitlists, RBAC, migrations, auditing, observability, and a shared registration-policy service.
- Flask
- SQLAlchemy
- PostgreSQL
- Alembic
- RBAC
- Docker
Real PostgreSQL final-seat concurrency test · atomic drop/promotion · OpenAPI/Swagger · Prometheus metrics · CI container build
06Database Systems · Team Project
Tax Payment Tracking System
Verified Spring 2024 CSIT 555 team-project history, paired with a clearly separated current Flask/SQLite reconstruction for tracking user-entered quarterly tax obligations and multiple partial payments with integer-cent monetary storage and derived status logic.
- Flask
- SQLite
- MySQL (historical)
- Docker Compose (historical)
- SQL
- pytest
Current repo: health + period/payment JSON APIs, cent-accurate money handling, relational constraints, focused tests · historical reporting/UI capabilities are not claimed as current implementation