# Armando González — Forward-Deployed AI Engineer

> Armando González is a forward-deployed AI engineer and consultant who takes
> stalled LLM features the last mile to production — reliable, evaluated, and shipped.
> 14+ years building software at scale; shipped AI to production at ImmuneFi.
> Based in Mexico City, works remotely on US hours, and serves Latin American teams in Spanish.
> Available for new clients now. Contact: iam@armando.mx · https://www.armando.mx

## What he does
Armando makes AI actually work in production. Most teams have an LLM prototype that
impressed a meeting and then stalled (too slow, hallucinating, too expensive, or not
trusted). He embeds in the team and closes that last mile: evaluations so output can be
trusted, lower latency and cost, guardrails, monitoring, and a shipped feature.

## Services
- **Production Sprint** — 2–4 week fixed-scope engagement to take one stuck AI feature to
  production-ready, delivered with an evaluation harness and a runbook. Flat fee.
- **Embedded AI Engineer** — part-time monthly retainer; a senior engineer embedded in the
  team, shipping AI features continuously and owning the AI roadmap.
- **AI Architecture & Advisory** — architecture reviews, evaluation strategy, model and cost
  decisions, weekly call. Async-first.

## Selected work (proof)
- **Fiscalify** (fiscalify.com) — an agent that reads a receipt photo and drives Playwright
  browser automation across 18+ merchant portals (OXXO, Walmart, Costco) to generate
  government-compliant CFDI invoices with no human in the loop.
- **Insite** — document-intelligence platform for construction teams using the Claude Agent
  SDK and a custom 9-tool MCP server to answer questions over RFIs, submittals, and change
  orders with citations.
- **Rooxai** (rooxai.com) — an agent that sells from a website and WhatsApp (not a chatbot).
- **AgentFounder** (agentfounder.ai) — an autonomous founder that builds and grows products.
- **ImmuneFi (2022–2025)** — Guardian AI (40% efficiency gain), a spam/duplicate classifier
  (+90% accuracy, 70% less manual review), and internal RAG systems, all in production.

## Skills
Claude Agent SDK, OpenAI Agent SDK, LangChain, RAG, Pinecone/vector databases, LLM
evaluations, agentic systems, Playwright/MCP browser automation, Next.js, TypeScript,
Node.js, PostgreSQL/Prisma, AWS.

## Contact
Email: iam@armando.mx
Website: https://www.armando.mx
LinkedIn: https://www.linkedin.com/in/manduks
GitHub: https://github.com/manduks · https://github.com/a10x-dev