Projects

Public demos, tools, and writing across AI security, private workflows, infrastructure, education, and small utilities.

Project Index

17 total projects 5 featured

Resume Contact GitHub

Hello.World Consulting

My consulting firm Featured

My consulting firm. Red teaming, architecture review, and secure deployment for teams shipping LLM systems.

Problem Clients were wiring LLMs into tools and data they couldn't afford to leak, usually after the integration was already live.
Outcome Cut exploit success and policy bypasses by 35% in tested systems. Local and private-cloud builds cut inference costs 20-45% in selected deployments.

AIConsultingSecurity

Open project

AI-Stats

Featured

Model prices, context windows, providers, latency, and benchmark signals in one workbench.

Problem Choosing a model meant fifteen open tabs: pricing pages, provider docs, benchmark leaderboards, none agreeing on format.
Outcome Model selection is a side-by-side comparison now, not a tab-hoarding exercise.

AI ModelsBenchmarksPricing

Open project

TRACED

Featured

A breach archive that reads like case files, plus a password-exposure check that never sends your password anywhere.

Problem Public breach data is everywhere and answers nothing. Disclosure posts are built for headlines, not for “was I in this?”
Outcome Breaches are browsable with dates and context; the password check keeps the secret on your machine.

SecurityBreach DataPrivacy

Open project

Semester Builder

Featured

A semester planner for students who want to see the whole degree, not guess at it.

Problem Degree requirements live in PDFs and advising portals, a bad place to plan four years of scheduling.
Outcome Semesters lay out visually, conflicts surface before registration, and plans export instead of living in screenshots.

PlanningToolingEducation

Open project

Internet Outage Atlas

Featured

An editorial archive of major internet outages: timelines, causes, and sources for each.

Problem Every big outage gets a postmortem and a news cycle, then both disappear. The patterns connecting them never get a shelf.
Outcome Failures sit side by side with dates and receipts, so recurring causes stop looking like one-offs.

InfrastructureOutagesAstro

Open project

AI-news

A curated feed of AI vendor updates: what actually changed this week.

Role Builder and curator

Outcome One feed to scan instead of a dozen vendor blogs on rotation.

CurationAIResearch

Open project

Dev Tools

Local-first browser tools for snippets, diagrams, SQLite, regex, schemas, tokens, IDs, JSON, and color checks.

Role Product engineer

Outcome Common inspection and conversion work stays in the browser instead of scattered across random utilities.

UtilitiesDeveloper Experience

Open project

Algorithmic Art

Generative coding experiments: visual systems that surprise their author.

Role Creative developer

Outcome A public gallery of repeatable generative methods, each one re-runnable.

Creative CodingGenerative Art

Open project

5 Whys

A toolkit for working through career decisions instead of circling them.

Role Builder

Outcome The reflection work has an address now.

CareerReflectionTools

Open project

RNG Chords

Random chord generator in the browser. Roll progressions until one lands.

Role Builder

Outcome Created a fast sketchpad for rolling harmony, previewing ideas, and exporting MIDI when something lands.

MusicMIDICreative Tools

Open project

AI-Drag Racing

Race AI providers head-to-head in the browser and watch the latency gap.

Role Product engineer

Outcome A drag-race format that makes latency and response differences visible in seconds.

AIBenchmarkingLatency

Open project

Prompt Info

Browser-based token counts, model-cost planning, tokenizer comparison, and agent-session budget inspection.

Role Builder

Outcome Built a browser tool for token analysis, model-cost estimates, planning receipts, and visible estimate limits.

LLM ToolsTokenizationCosting

Open project

PoliBench

Benchmark of how AI assistants answer political questions. Every release ships with its evidence.

Role Research systems builder

Outcome Built a public benchmark dashboard with model profiles, methodology, and release evidence.

AI EvaluationBenchmarksResearch

Open project

NautilusBot

Project site for a local-first dictation, transcription, and meeting-audit app.

Role Product engineer

Outcome Published the product surface for recording, transcription, local AI analysis, and evidence export.

TranscriptionLocal AIProduct

Open project

Field Viewer

Client-side physics sketch for inspecting electron and photon field packets.

Role Builder

Outcome Built a browser simulation for packet motion, antiparticles, and conservation checks.

PhysicsCanvasSimulation

Open project

Science Wonders

Interactive science reference for accelerators, detectors, and gravitational-wave observatories.

Role Builder

Outcome Created an educational site with local TypeScript content, interactive diagrams, and small simulations.

ScienceEducationInteractive

Open project

ZenType

Typing practice without the racetrack.

Role Product engineer

Outcome A practice loop quiet enough that you actually return to it.

TypingUXMindfulness

Open project