Independent Projects
Independent projects
Products and experiments I conceived, designed, and built independently, outside of employer work. Work done for employers is described under experience, and published research under writing.
AI Product · Independently Created
SignalCheck — A Trust Score for What You’re Reading
A browser extension and API that combine domain reputation with AI-generated-content detection into a single, explainable trust score. Built end to end, outside of any employer.
- Role
- Product · Design · Builder
- Context
- Built end to end, outside of any employer
- Status
- v3 in development · public repository
The problem
Domain reputation and AI-content detection answer different questions and live in different tools. A reader deciding whether to trust a page has to hold both in their head, and neither signal means much alone.
My role
Product · Design · Builder. I framed the problem, designed the scoring model and system boundaries, and built it. Independent work, outside of any employer.
What it does
- Manifest V3 Chrome extension with a FastAPI backend.
- Fuses Google Safe Browsing reputation with Gemini AI-content detection into one 0–100 Signal Trust Score.
- The score surfaces inline as the reader browses, with the contributing signals shown separately rather than hidden behind the number.
- An admin view tracks scoring behavior across requests, which is how the weighting gets tuned.
What it deliberately does not do
- The score prompts a closer look. It is not a verdict, and it is not a moderation decision.
- AI-content detection is probabilistic. It is never presented as proof of authorship.
- Low-confidence results display as low-confidence rather than rounding to a clean number.
- Where reputation and content signals disagree, the disagreement is shown rather than averaged away.
Outcome
A working prototype and a public repository, plus a written account of how the score is composed, where the two signals disagree, and which judgments stay with the reader.