ProjectHub Scout is online 100% free stack

Meet Scout, my AI recruiter assistant

Scout helps visitors ask questions about my projects, background, skills, blogs, resume details, and career fit without digging through every page manually. It runs entirely on free tiers: GitHub Pages, a GCP VM, free LLM providers, and a local Ollama fallback.

Bradley Matera · Recruiter assistant

Calm, concise, honest, and grounded in my real project data — with a zero-cost multi-provider LLM network behind it.

What this is

ProjectHub is a vanilla JavaScript, no-build chat widget that can be embedded on a site with one script tag. It showcases my web development projects, CodePens, GitHub work, recruiter information, and portfolio data through a floating assistant named Scout.

The goal is not a hard-coded FAQ bot. Scout retrieves from real project data, recruiter knowledge, source material, and a free multi-provider LLM network so it can answer new questions honestly instead of repeating canned responses.

Current status

Frontend GitHub Pages widget

Loaded through the public ProjectHub script. No build step, no hosting cost.

Backend Free multi-provider LLM router

Routes open-ended questions through Groq, Cloudflare Workers AI, GitHub Models, Gemini, and a local Ollama fallback on a GCP VM.

AI Grounded-first RAG

Deterministic answers for facts; the free provider network adds natural phrasing when the data supports it.

The current direction is a zero-cost recruiter assistant that uses the resources already available: GitHub Pages, ProjectHub data files, a free GCP VM tier, free LLM providers, local Ollama models, and structured recruiter knowledge.

What Scout should answer

  • Project questions: Explain my projects, CodePens, GitHub repos, tech stack, and what each project proves.
  • Recruiter questions: Summarize my skills, background, certifications, AWS internship, education, and job fit.
  • Career-fit questions: Compare my real experience against roles like web developer, IT support, cloud support, DevOps, technical writing, QA, or data-focused work.
  • Grounded answers: Use real data from ProjectHub, recruiter knowledge, source material, blogs, pages, and resume guardrails.
  • Honest limits: If the data does not support an answer, Scout says that instead of inventing experience.

Architecture

Recruiter Hub page
→
ProjectHub widget
→
Free LLM router
→
ProjectHub data + RAG
→
Scout response

Scout first checks whether a question can be answered deterministically from the knowledge JSON. For open-ended questions, it sends the verified context through a priority network of free providers (Groq, Cloudflare Workers AI, GitHub Models, Gemini) and falls back to a local Ollama model if none succeed. Every generative reply is validated against the source facts before it is returned.

Free stack breakdown

GitHub Pages

Hosts the widget, landing page, and knowledge JSON for free.

GCP VM (free tier)

Runs the Node.js chat API and local Ollama fallback without a monthly bill.

Free LLM providers

Groq, Cloudflare Workers AI, GitHub Models, and Gemini free tiers rotate automatically.

Local Ollama

Final fallback using small open-weight models (smollm2:135m) so the assistant is never hard-down.

How to embed ProjectHub

This is the script tag used to load the floating ProjectHub chat widget: 

<script src="https://bradleymatera.github.io/ProjectHub/ProjectHub.js?v=9"></script>

If the script is cached aggressively, bump the version query like ?v=9.

Important links