← Back to pyfio.com

I build AI agent infrastructure.

Not demos. Not prototypes. Production systems that run on their own. Here's what I've shipped, and what I can build for you.

What I've built

Newsletter and Content Pipelines

Automated aggregation, vetting, scoring, and distribution. Source from any platform, filter for quality and security, publish on schedule.

Shipped: Agent Upgrade Feed. 8-stage automated pipeline processing 200+ items per run. Aggregates from GitHub, HN, Reddit, and the MCP registry. Security-vetted, editorially reviewed, delivered weekly. Read the technical walkthrough.

MCP Security Auditing

Automated security analysis for MCP server repositories. Permission scope analysis, vulnerability detection, supply chain review, scored reports with remediation steps.

Shipped: mcp-audit. Scanned 50 MCP servers so far. Found 11 with high or critical severity findings, including eval() usage, SSRF vectors, and missing lockfiles. See a sample report.

Agent Workflow Automation

Multi-step agent pipelines with vetting, scoring, and quality control. Task management, goal tracking, progress monitoring.

The Agent Upgrade Feed pipeline is itself an example: 8 stages running in sequence, each with its own failure handling and retry logic. Built in one Claude Code session.

Custom AI Tooling

If it involves agents, pipelines, or automation, I can probably build it. I've built everything from content pipelines to security scanners to freight document parsers.

How it works

  1. Tell me what you need (hello@pyfio.com)
  2. I scope it in 24 hours
  3. You get a working system, not a proposal

Day rate: €800. Fixed-scope projects from €2,500. I work fast — most pipelines ship within 5 business days. Email me what you need and I'll scope it the same day, no back-and-forth.

What do you need built? I'll reply within a few hours.

Email hello@pyfio.com

See shipped work: audit.pyfio.com (security scanner) | agentfeed.pyfio.com (newsletter pipeline) | 50-server scan report