📍 Santiago, Chile | Software Engineer | System Design
Fullstack Engineer focused on AI-driven systems, platform engineering, and scalable architectures. I build reliable systems, integrate AI into operational workflows, and deliver measurable improvements in performance and observability.
I keep my development stack small, TUI-friendly, and focused on agentic workflows. Lately I am writing less glue code by hand and leaning more on structured specs, review loops, and tools like Hex when they help me think faster.
- LLMs: GPT and GLM as my main model families.
- Agent harnesses: Codex and OpenCode for repo exploration, implementation, reviews, and verification.
- Terminal: Ghostty with TUI-first tools and local scripts.
- Editor: Visual Studio Code for code review, navigation, and focused edits.
- Production: Railway, Vercel and Supabase for shipping apps and services quickly.
- Workflow frameworks: Spec Kit for spec-driven development, Get Shit Done for context engineering and execution flow, and Skills for reusable AI skill patterns.
- Most used skills:
grill-me,to-prd,tdd,improve-codebase-architecture,vercel-react-best-practicesandfrontend-design.
A personal knowledge system that captures, curates, and connects everything I learn — powered by AI agents and Obsidian.
| Project | Description |
|---|---|
| Second Brain Ecosystem | Bilingual guides + templates to set up your own Second Brain from scratch with Obsidian, with an optional AI layer |
| Librarian | AI agent that maintains my Obsidian wiki — reads raw sources, classifies, deduplicates, and generates curated wiki pages (proposal-first, human-approved) |
| Researcher | Agentic web search agent (Search-o1 pattern) — finds and synthesizes information missing from my knowledge base |
| Content Toolkit | Media ingestion tools — YouTube → transcription → smart summary → Obsidian raw/ |
| Smart Summarize | 9 evidence-based summarization methods for AI agents (Feynman, SQ3R, Cornell, Adler, and more) |
- Mission Control — Open-source agent orchestration framework. Spec-driven development with AI-powered agents.
- SmartOps AI — AI Decision & Automation Platform. Analyzes operational events with LLMs and executes automated responses.
- AI Code Guardian — AI-powered linting tool for PRs. Automated code review and validation.
- Club de Barrio — Fullstack dashboard para clubes de fútbol.
- Van's Office — Personal CLI workflow toolkit.
Walmart Chile — Software Engineer (2021 – 2025)
- Migrated legacy search to unified platform using Go + Algolia
- Built Inventory Management System & Promise Engine
- Modernized services from JavaScript → Java 8
- Observability and resilience with Dynatrace, Grafana, Kubernetes
- Migrated secrets from Vault → Akeyless
Telefónica — Software Developer (2017 – 2021)
- Built web platforms with Angular/TypeScript/Node.js and GCP
- Mobile App Mi Movistar (Flutter)
- Trained teams on Clean Architecture and Clean Code
Core Skills & Current Stack
Languages: TypeScript, JavaScript, Go, Java, Python
Backend: NestJS, Node.js, Spring Boot, Hono
Frontend: React, Angular, Next.js
Cloud & Infra: Kubernetes, Docker, GCP, Azure
Databases: PostgreSQL, MongoDB, Redis, MariaDB
AI & Tools: OpenAI, LLMs, Agent patterns, Kafka, Algolia, Dynatrace, Grafana
Agentic development: Codex, Claude, prompt notes, review loops
Daily tools: Git, terminal, VS Code/Cursor, local scripts
Focus: minimal tooling, explicit context, safe iteration, measurable improvements
Communication is the core skill of this moment.
A minimal workflow keeps the signal high, the context clear, and the work moving.
Random Facts
- Completing my list of 100 analog hobbies.
- Photography and color hunting.
- Gym to release stress.