NBA Shot Data Engineering Package
SQLite-backed multi-source shot modeling and archetype workflow.
Build Notes
- Merges ESPN and NBA Stats shot data across seasons
- Applies schema validation before feature generation
- Uses rerunnable SQLite upserts for stable refresh behavior
- Builds xFG, residual, and SDI feature layers for modeling
- Automates daily updates through launchd scheduling
Tech Focus
R
Python
SQLite
PyGAM
scikit-learn
Streamlit
AI Multitool Assistant
Two-service AI web product with secure auth and tool orchestration.
Build Notes
- React/Vite frontend paired with Django REST backend
- JWT token lifecycle with refresh and user-scoped storage
- Gemini + LlamaIndex ReAct tool orchestration
- PDF upload/index/query flow for document-grounded responses
- Tool modules for market data, weather, and news
Tech Focus
React
Vite
Django REST
JWT
Gemini
LlamaIndex
ESPN NBA Data Pipeline
R package with parallel collection and schema-safe upsert loading.
Build Notes
- Parallel collection wrappers with retry and rate-limit handling
- Schema maps enforce stable typed outputs
- Composite-key upsert strategy for idempotent loads
- Manifest checks verify schedule completeness
- Fixture-based testthat suite covers parse and DB behavior
Tech Focus
R
httr2
DBI
RSQLite
future
testthat
NAU Course Catalog Scraper
Selenium + PDF prefix extraction + CSV/report analysis pipeline.
Build Notes
- Builds prefix inventory from PDF and crawl logs
- Supports resumable Selenium crawl with term-aware outputs
- Runs precision, broad, and ethics analysis scripts separately
- Exports clean CSV artifacts for downstream reporting
- Publishes findings in R Markdown report outputs
Tech Focus
Python
Selenium
pdfplumber
pandas
thefuzz
R Markdown
NBA Win Probability Models
Leakage-aware modeling with rolling features and reproducible comparisons.
Build Notes
- Builds lagged rolling metrics and matchup differentials
- Compares ridge/elastic/logistic/random forest pipelines
- Integrates Elo variants as predictive signals
- Uses chronological splits to preserve real-world ordering
- Tracks AUC and calibration tradeoffs by model
Tech Focus
R
glmnet
pROC
randomForest
slider
hoopR
Deep Learning Projects
Progressive deep-learning coursework with reproducible experiment tracking.
Build Notes
- ANN baseline modules for data, loss, and training
- HPO loops with grid/random strategies and W&B logging
- CNN interpretation through filter visualization tooling
- RNN sequence modeling with custom tokenizer utilities
- Shared seed controls for reproducibility across projects
Tech Focus
PyTorch
TorchVision
W&B
pytest
Conda
NBA Analytics Platform
Production monorepo with secure API, billing, and high-performance frontend.
Build Notes
- Fastify API modules split across auth, billing, Q&A, and routes
- Regex intent mapping prevents direct user SQL execution
- Stripe checkout and webhook flow manages subscriptions
- Zod validation, JWT policy, rate limiting, and sanitization hardening
- Redis and cache-layer optimizations reduce response latency
Tech Focus
TypeScript
Fastify
Next.js
Prisma
PostgreSQL
Redis
NAU Capstone: Sports Expected Points Analysis
Cross-sport expected-points modeling and interpretability toolkit.
Build Notes
- xFG/xG modeling scripts for NBA and NHL datasets
- Calibration diagnostics including Brier and reliability curves
- GMM archetype modeling with PCA visualization
- Value analysis combining POE with salary data
- Interactive Streamlit app for presenting capstone findings
Tech Focus
Python
scikit-learn
PyGAM
Streamlit
plotly
matplotlib
NBA Dockerized Scrape Pipeline
Dockerized PostgreSQL ingestion workflow with idempotent upserts.
Build Notes
- Parallel fetching stage built with concurrent.futures
- Normalization maps ESPN JSON into relational model tables
- Chunked SQLAlchemy upserts enforce idempotent writes
- nba_ingest_failures table captures pipeline breakpoints
- Makefile commands orchestrate ingest and infrastructure lifecycle
Tech Focus
Python
Docker
PostgreSQL
SQLAlchemy
Makefile
Stuxnet Cyberwarfare Analysis
Technical malware research and publication-quality security reporting.
Build Notes
- Breaks down propagation via USB/LNK exploit chain
- Analyzes four zero-day exploits and privilege escalation paths
- Documents PLC targeting logic and operational manipulation
- Compares impact context with major cyber incidents
- Publishes full report using R Markdown + LaTeX formatting
Tech Focus
R Markdown
LaTeX
ICS/SCADA
Exploit analysis
Technical writing