Full-stack developer crafting AI platforms, developer tools, and products that solve real problems.
I'm a full-stack developer with a passion for turning complex problems into elegant, working software. From AI agent platforms to satellite imagery tools for farmers, I build across the entire stack — Go, Rails, Elixir, Next.js, and Python.
My flagship project, Flukebase, is an AI-first agent platform that serves as the backbone for all my projects — managing infrastructure, deployments, DNS, and AI agent orchestration from a single Go monorepo.
I believe in building with purpose: every project in my portfolio addresses a real problem I've encountered or a gap I've identified. Whether it's making satellite data accessible to Brazilian farmers, creating voice tools for Linux, or building emotional wellness tracking — the thread that connects all my work is making complex systems approachable.
Each project starts with a real problem. Hover to see the thinking behind the work — the problem, the approach, and the impact.
AI-first agent platform
A Go monorepo for building, deploying, and orchestrating autonomous ARIA agents with effect-based architecture. The backbone that manages infrastructure, CI/CD, DNS, email, and deployments for all my projects.
Problem
Managing multiple projects across different servers, domains, and deployment pipelines was a fragmented, manual process. No single platform could orchestrate AI agents alongside traditional infrastructure.
Solution
Built a unified Go platform with MCP server integration, agent orchestration, and full infrastructure management — from DNS to SSL to deployments. Every project in this portfolio is managed through Flukebase.
Impact
Manages 8+ domains, 5+ deployed apps, and orchestrates AI agents across all projects. Single control plane for the entire portfolio.
Linux-first AI dictation assistant
A voice writing assistant with pluggable ASR engines, evaluation-driven development, and AI text polishing. Built for Wayland with local-first processing.
Problem
Linux lacks a reliable, high-quality dictation tool. Existing solutions are Windows/Mac-centric, tightly coupled to single ASR engines, and have no way to evaluate transcription quality systematically.
Solution
Created a pluggable architecture with multiple ASR engine adapters (Whisper, Deepgram, etc.), an evaluation-first development pipeline, and Wayland-native text insertion with fallback strategies.
Impact
Evaluation-first approach surfaced quality issues 3x faster than intuition-driven development. Supports multiple ASR backends through a common adapter interface.
Emotional wellness tracking
A wellness tracking app that helps users understand their emotional patterns through daily check-ins, mood visualization, and trend analysis.
Problem
People struggle to understand their emotional patterns. Traditional journaling is inconsistent, and existing mood apps lack depth or actionable insights.
Solution
Built a Rails 8.1 app with Hotwire for real-time interactions, rich mood visualizations, and pattern recognition that surfaces emotional trends over time.
Impact
Live at feeltrack.tech with SSL. Full production deployment managed through Flukebase infrastructure.
Satellite monitoring for Brazilian farmers
Next.js application enabling Brazilian farmers to monitor land parcels via satellite imagery from CBERS and Sentinel-2 satellites, with SICAR WFS integration for parcel boundaries.
Problem
Small and medium Brazilian farmers have no affordable way to monitor crop health, deforestation, or land changes using satellite data. Existing tools are expensive and complex.
Solution
Built a Next.js app that integrates CBERS/Sentinel-2 satellite imagery with SICAR (Brazilian Rural Environmental Registry) WFS data, providing accessible land monitoring with parcel overlay visualization.
Impact
Makes satellite-based land monitoring accessible to farmers who previously had no visibility into their parcels from space.
Agentic UX for Pilates studios
An agentic UX platform that automates and enhances the business journey for Pilates studio owners, employees, and business partners through intelligent agent-driven interactions.
Problem
Pilates studio owners juggle scheduling, client management, payments, and marketing across disconnected tools. The operational overhead limits their ability to focus on instruction.
Solution
Designed an agent-driven platform where AI handles scheduling optimization, client communication, business analytics, and partner coordination — letting studio owners focus on what they do best.
Impact
Comprehensive business automation platform designed for the growing Pilates industry.
Food surplus marketplace
Connecting consumers with restaurants, bakeries, and grocery stores to rescue surplus food through a dual offer model with geolocation-based discovery.
AI hardware engineering assistant
An AI-powered assistant for hardware engineers with chat, requirements generation, pricing intelligence, and inventory management capabilities.
AI-powered learning platform
An AI-powered learning platform for the 2026-2030 cycle, featuring YouTube and Mux video integration, pgvector for semantic search, and adaptive learning paths.
AI terminology reference
A comprehensive AI glossary with an Elixir Phoenix API and Python ML pipeline, using pgvector for semantic embeddings and multi-service Docker deployment.
Founder workflow platform
A workflow platform for founders with Clerk authentication, Stripe payments, and end-to-end Playwright testing.
Court assistant for California
A California small claims court assistant with guided workflows, form generation, and comprehensive E2E testing.
Travel itinerary app
A travel itinerary planning application that helps travelers organize and optimize their trip plans.
I pick tools based on the problem, not the trend. From Go's performance for agent orchestration to Rails' productivity for web apps — each project uses what fits best.
Have a project idea, want to collaborate, or just want to say hi?