I build agentic systems and the infrastructure that keeps them trustworthy.
Founder & Principal Engineer at VeriSwarm — the cross-platform trust layer for AI agents: deterministic 5-dimension trust scoring (including Brier-scored confidence calibration), transparent MCP tool-call interception, prompt-injection defense, and tamper-evident audit trails. Built solo, running in production on a self-hosted HA Kubernetes cluster.
- veriswarm-sdk — Node and Python SDKs plus a 90+ tool MCP server with Claude Code, Gemini CLI, and Codex integration (automatic PII-guarding hooks included)
- APEX — multi-agent orchestration on the Claude Agent SDK: planning, implementation, testing, and review agents with Git worktree parallelism and a real-time web UI
- AI Infrastructure Curriculum — a 13-role curriculum maintained by autonomous agent runners with hard guardrails, append-only audit logs, and GitHub triage automation
- Streamspace — open-source multi-user container streaming platform (Go)
- Terminal Velocity — multiplayer space trading game playable entirely over SSH (Go)
Agentic AI: Model Context Protocol (MCP) · Claude Agent SDK · Claude Code plugins/hooks · LangChain · prompt-injection defense · RAG · LLM routing
15 years of platform engineering before the agents showed up — cloud automation across 7 global data centers (~7,000 VMs, ~400 enterprise customers), Cloud Center of Excellence work, and more Terraform than I care to admit.
📍 Phoenix, AZ · 📧 contact@joshua-ferguson.com · 💼 linkedin.com/in/joshuaaferguson · 📄 resume.joshua-ferguson.com




