Products, research tools, and smaller builds. Watch the recorded demos, open the verified live sites, or continue to each project’s GitHub repository.
ParableParable is an open-source catalog of 88 React components and 78 complete templates that developers can copy into their own projects.
I built the Next.js catalog, live previews, shadcn-compatible registry, documentation, and stateless MCP endpoint.
Forge AIForge AI is a self-hostable Next.js workspace for turning application briefs and targets into editable websites inside restricted development environments.
I built the verified workspace and the recorded local workflow for brief entry, design selection, generation, preview, editing, and export.
SpecterSpecter is a native macOS terminal that runs real login shells through POSIX pseudoterminals and draws its terminal grid with Core Text and Metal.
I built the PTY lifecycle, VT parser, bounded screen state, Metal renderer, native workspaces, themes, companions, and validation suite.
MosaicMosaic is an iPhone app where privately verified acts add equal-size tiles to a shared artwork.
I built the SwiftUI participant and organizer flows, Supabase backend, RLS policies, private media handling, widgets, and final reveal.
BBHUNTER is an authorized security-research toolkit with 13 reconnaissance phases, 58 hunt skills, and independent validation.
I built the CLI orchestration, report-derived knowledge base, sibling predictor, MCP services, and evidence workflow that separates candidates from verified findings.
FaceCheck Pro is a limited educational face-retrieval study using 512-dimensional ArcFace embeddings and a 252-face demonstration index.
I built the Next.js upload interface and FastAPI query path, kept uploads in memory, and separated browser feedback from server-side matching.
Formwork is a collection of 27 standalone HTML, CSS, and JavaScript templates for graphics, motion, sound, and interaction.
I built each template without a build step and documented the main interaction so every example can be opened and studied on its own.
Image Classifier is an Iowa State project comparing a custom CNN with MobileNetV2 across five mushroom species.
I built the notebook, data augmentation, evaluation, and result visualizations; MobileNetV2 reached about 80 percent validation accuracy on the documented split.
Sentiment Analyzer is an Iowa State project organizing 75 sample restaurant reviews by sentiment and topic.
I built the single-page interface and Chart.js visualizations using the documented Excel, ChatGPT, and Claude classification workflow.
