An OpenCode plugin that tracks real-time model metrics, token usage, and session costs directly within the TUI sidebar with cross-restart persistence.
Hi, my name is
Alonso Santana
I build high-performance software.
Full-stack engineer building fast web applications, AI agent workflows, and automated CI/CD pipelines with TypeScript.
01.About
Hello, I'm Alonso Santana. I'm a full-stack engineer driven by a fundamental question: how does this actually work under the hood? That deep technical curiosity shapes my approach to software—I don't just assemble frameworks, I evaluate trade-offs. I focus on system architecture, stack selection, and long-term scalability to build fast web applications backed by resilient CI/CD pipelines and clear infrastructure.
Lately, I've been diving into modern AI engineering, designing agentic workflows with LangGraph, telemetry with Langfuse, and custom tools with the Model Context Protocol (MCP). I write about my architectural experiments and what I learn along the way at Code Habits Blog.
A few technologies I work with:
- TypeScript
- React
- Vite
- Astro
- Node.js
- Supabase
- LangGraph
- Docker
- GitHub Actions
- Linux/Zsh
02.Experience
Full-Stack Engineer Fintech Platform
May 2023 - Jun 2026
- Engineered and optimized an internal management dashboard serving over 10,000 users, integrating automated KYC verification and digital signature workflows to streamline onboarding.
- Transitioned to a hybrid DevOps and Full-Stack role, managing cloud infrastructure across AWS (EC2, S3, RDS) and automating deployment pipelines using GitHub Actions.
- Refactored core Node.js backend services and MySQL query architectures, drastically improving execution performance and data throughput while maintaining 100% system stability.
- Developed end-to-end web interfaces and APIs using Next.js, TypeScript, and Node.js for financial and user-management systems.
03.Blog
Terminal and Command Line: my notes from Arch and Zsh
A practical guide to the terminal, Bash, Zsh, essential commands, redirections, permissions, processes, and modern CLI tools.
04.Projects
An MCP server that helps LLMs select optimal AI models for coding agents by cross-referencing OpenCode subscription data with OpenRouter benchmarks.
codiceagents
An experimental project exploring multi-agent architectures and deterministic state workflows using LangGraph, paired with Langfuse for full execution tracing.