I build automations that keep running after the demo. Almost 3 years building software, the last two on automation and AI, currently full-time in software and automation. Freelance on the side.
WHAT I BUILD
• n8n and Make workflows, plus custom code where low-code runs out — usually at retries, rate limits, or anything stateful
• AI agents with Claude and the OpenAI API, including MCP servers
• API, CRM and ERP integrations, webhooks, and the alerting layer that tells you when one stops silently
• Python, TypeScript, C#/.NET, SQL
WHY MY BUILDS SURVIVE
Most automations do not fail loudly. An API returns 429, a webhook fires twice, a run stops and nobody notices for a week. Before I build anything you get the failure cases in writing: what happens when the source is down, when the data is malformed, when a run repeats. Then I build those first.
PROOF, NOT ADJECTIVES
GitHub - andresFLZ/mcp-trm-colombia: MCP server for Colombia's official USD/COP exchange rate (TRM) · GitHub — an MCP server with 31 tests, CI passing on three Python versions, and a recorded demo of it rejecting a bad request instead of inventing an answer. Read the code before you decide.
ENGAGEMENT
Fixed-scope builds from $150, or $15/hour. I am new to the freelance platforms and priced accordingly while I build a track record — the work is the same work. Handover always includes the workflow JSON and documentation good enough that someone else can maintain it.
Tell me what triggers the process today and what should exist when it finishes, and I will tell you honestly whether automating it is worth it. If it is not a fit for my stack, I will say so.