Selected work showing how I reduce risk in cloud, security, infrastructure, and AI-enabled workflows.
I help teams build systems that are safer, more legible, and more useful when AI enters the workflow.
AI/MCP security integration
Built a read-only MCP server that lets AI agents query Cisco Firepower Management Center without handing them unsafe write paths.
Value: Shows how to connect LLM workflows to real security tooling with bounded permissions, token lifecycle handling, pagination, rate limiting, tests, typing, and linting.
Proof: Public GitHub repository; designed around Cisco FMC API realities like short-lived tokens and refresh limits.
Cloud systems under operating constraints
Worked close to large production cloud environments, support paths, migrations, managed-service behavior, and customer-facing operating constraints.
Value: Useful for reviews where IAM, networking, logging, quotas, managed-service assumptions, or rollback paths matter more than the architecture diagram.
Proof: Practical work across VPCs, load balancing, DNS, private connectivity, streaming/search services, operational reviews, incident preparation, and migration planning.
Network infrastructure under constraint
Pre-simulated a redundant Point of Presence from Cape Town using GNS3 and VMware, then deployed it in a mobile-network data center in 72 hours with no prior physical access.
Value: Demonstrates how Leon reduces risk before execution: model the system, test the assumptions, then ship under real constraints.
Proof: Redundant PoP delivered in three days after remote simulation.
AI-first professional presence
Built this site so humans and agents read the same underlying public facts through HTML, JSON, llms.txt, well-known metadata, and MCP.
Value: Shows the AI-first component directly: machine-readable identity, representation instructions, and agent-accessible context without burying the human visitor in theory.
Proof: Public endpoints include /api/v1/profile.json, /api/v1/offers.json, /api/v1/case_studies.json, /llms.txt, /.well-known/agent-card.json, and /mcp/.
released
Read-only MCP integration for Cisco Firepower Management Center. It gives AI agents useful security context while keeping write paths out of scope.
Proof markers: Automatic token management; Rate limiting and concurrency control; Transparent pagination; Pytest, mypy, and ruff quality gates.
active development
Personal AI engineering partner and governance proving ground. The public value is not the name; it is the pattern: bounded tools, persistent context, review loops, and explicit operating constraints.
active development
Compliance-readiness workflow tool for small businesses, framed around scoped evidence, no-CUI boundaries, review gates, report-language restraint, and production auth rather than certification promises.
Proof markers: No-CUI guardrails and plain-language warnings; Workflow gates before reports; Report wording that avoids certification overclaims; Access-controlled production deployment pattern.
active
AI-first professional presence that exposes the same facts to humans and agents through HTML, JSON, llms.txt, agent metadata, and MCP.
Proof marker: Profile-as-code without making the human buyer read a manifesto.
Start with the services page if you want a scoped offer, or contact me directly if the problem is still messy and needs shaping.