New Year, New You Portfolio Challenge - Evolving "Portfolios" into an Agent

This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI

About Me

I'm Michael Weed, a Solution Architect and Agentic AI Engineer who's spent the last decade bridging enterprise systems (Salesforce, AWS, Google Cloud) in high-compliance environments (HIPAA, PCI-DSS, SOC2). I've grown tired of static portfolios that say "I know AI" without proving it.

So I built Atlas-G Protocol, a portfolio that isn't just about AI - it is AI. Give it a try below.

Portfolio

How I Built It

The Core Idea is Agent-to-Agent Communication
Traditional portfolios are read by humans. Mine is designed to be queried by humans and agents. Built with the power of Google Antigravity to use Cloud Run functions.

The Atlas-G Protocol exposes an MCP (Model Context Protocol) server that allows any MCP-compatible environment - Claude Desktop, Gemini CLI, Cursor, or Antigravity - to connect and interview me programmatically. Your agent can ask my agent about my experience, verify my claims, and audit my project architectures, all through standardized tool calls.

This is the future of recruiting: Agents screening agents. Instead of a recruiter scanning a PDF, their AI assistant calls
mcp_query_resume("Experience with HIPAA-compliant systems")
and gets a verified, structured response.

Tech Stack

https://atlas-g-protocol-651633010401.us-central1.run.app/
The function is also deployed to full domain michaelweed.xyz

The Governance Layer

Every response passes through a validation layer that:

1. Cross-references claims against my resume knowledge graph
2. Blocks hallucinations before they reach the user
3. Logs compliance checks in real-time to a visible audit stream
4. If the agent tries to claim I worked somewhere I didn't, the Governance Layer catches it.

This is the kind of production-grade guardrail I architect for my clients, now demonstrated live.

What I'm Most Proud Of

1. The "Interview Me" MCP Server Other developers can point their AI tools at my portfolio and have a conversation with it. Here's how:

For Cursor / Claude Desktop / Antigravity users, the easiest way is to add it as a Native SSE server:

• Type: SSE
• URL: https://atlas-g-protocol-651633010401.us-central1.run.app/mcp/ Alternatively, use the mcp-remote bridge:

{
  "mcpServers": {
    "atlas-g-protocol": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://atlas-g-protocol-651633010401.a.run.app/mcp"]
    }
  }
}

Then your agent can call tools like:

mcp_query_resume("Salesforce experience")
→ Semantic search over my career
mcp_verify_employment("Salesforce", "Solution Architect")
→ Verify my claims
mcp_audit_project("Atlas Engine")
→ Deep-dive architecture with GitHub links
mcp_check_availability()
→ See if I'm available for contract work

1. The Live Audit Log
   Watch the agent "think" in real-time. Every compliance check, every validation step, every governance decision is streamed to the UI. It's not a gimmick—it's proof of work.

2. Privacy by Design
   My resume data never touches the public repo. The
   resume.txt file is gitignored, and the template is provided for anyone who wants to fork and personalize. The agent reads from local data, so my PII stays private while my capabilities stay public.

Try it now: Connect your agent to my portfolio and see what agentic architecture actually looks like in production.