Muhammad Hamid.
AI Automation Engineer

I build agentic systems that survive contact with production.

MCP servers, evaluation harnesses, execution traces and permission gates — wired into real business systems through APIs, webhooks and Python services. Eight years in growth engineering and technical SEO, now applied to making AI agents verifiable.

28open-source repositories
61MCP tools shipped in shopify-growth-mcp
50merged pull requests across 25 upstream projects

See the workGitHub

What I build

Agents with governance

Tool permissions, approval steps, execution traces, cost ceilings, and eval harnesses that measure accuracy before anything ships.

MCP servers and integrations

Model Context Protocol servers that expose real business systems — commerce, analytics, search data — to Claude, Cursor, VS Code and Codex.

Python automation

FastAPI services, scheduled pipelines and CLI tools that replace recurring manual work with deterministic, testable code.

Technical SEO and measurement

Crawlers, audit engines and conversion-tracking diagnostics that turn ambiguous marketing data into specific fixes.

Selected work

MCP · TypeScript

shopify-growth-mcp

61-tool MCP server for Shopify operations: SEO and GEO audits, deterministic analytics, inventory intelligence, feed quality checks, and approval-gated marketing workflows.

Eval · Python

ai-agent-testing-lab

Reliability and evaluation harness for agents: labelled suites, CI integration, and accuracy, calibration, cost and latency reporting.

In design

mcp-conformance

A conformance and security scanner for MCP servers: protocol checks, tool-schema validation, description-injection rules, context-cost reporting, SARIF output and a timestamped badge. Spec is public in the project notes.

How I work

Every claim on this site is backed by a repository you can read, and every tool I ship has to pass its own checks before it ships. When measurement is impossible, I say so instead of inventing a number.