Staff Software Engineer with 20 years of remote backend experience. I ship production AI tooling across three runtimes — Ruby (Rails), Elixir/Phoenix, and Rust — using MCP, ReAct, and disciplined TDD/DDD to make AI actually useful in production codebases.

I build AI tooling and high-throughput backend systems across Ruby (Rails), Elixir/Phoenix, and Rust. My focus is making AI useful, safe, and measurable in production: MCP servers, context bridges, evaluation sandboxes, and memory-safe CLIs.

• Ruby
  • rails-ai-bridge — 5,846 RubyGems downloads
  • ruby-skill-bench — 1,734 RubyGems downloads
  • rails-agent-skills — 22 GitHub stars
• Rust
  • brigid — 169 brigid-core crate downloads; turns codebases into LLM-generated tutorials
  • rs-guard — 242 rs-guard crate downloads; multi-provider AI code-review CLI
• Elixir
  • elixir-phoenix-skills — 47 agent skills for idiomatic Phoenix/LiveView/Ecto

• Polyglot depth: real, shipped code in three different runtimes, not toy repos.
• AI/tooling credibility: MCP servers, eval frameworks, and safe Rust CLIs — not "vibe coding".
• Production impact: 5,846 + 1,734 RubyGems downloads, 242 + 169 crate downloads, and a 20-year track record scaling systems past 10k TPS.

flowchart LR
    subgraph "Skill Ecosystem"
      Rails[Developer with Rails app] --> Bridge[rails-ai-bridge<br/>Rails context + MCP]
      Bridge --> IDEs[IDEs: Cursor, Claude, Copilot, Windsurf, Gemini]
      Bridge --> Bench[ruby-skill-bench<br/>Eval / ROI of context]
      Codebase[Codebase] --> Brigid[brigid<br/>LLM tutorial engine]
      Codebase --> Guard[rs-guard<br/>AI code review]
      Phoenix[Phoenix/LiveView app] --> Skills[elixir-phoenix-skills<br/>47 agent skills]
    end

• rails-ai-bridge — Ruby. Zero-config MCP server + context files for Rails. 5,846 RubyGems downloads, 94.49% test coverage.
• ruby-skill-bench — Ruby. Evaluation engine that measures the ROI of AI context. 1,734 RubyGems downloads, multi-provider, blind LLM judging.
• rails-agent-skills — Ruby. 28 Rails-specific skills + 9 workflow templates. 22 GitHub stars.
• brigid — Rust. Turn any codebase into an LLM-generated tutorial. brigid-core: 169 crate downloads.
• rs-guard — Rust. Multi-provider AI code-review CLI (DeepSeek, Qwen, Kimi, OpenRouter, OpenAI). rs-guard: 242 crate downloads.
• elixir-phoenix-skills — Elixir. 47 agent skills for idiomatic Phoenix/LiveView/Ecto.

Ruby / Rails	Elixir / Phoenix	Rust	AI Engineering	Systems & Architecture

• Dealerware (Software Technical Lead | May 2022 — April 2026): Originally joined as a contractor via 3Pillar Global, hired directly and promoted from Mid-level to Senior, and subsequently to Software Technical Lead due to high performance. Directed cross-functional distributed squads across 20+ production codebases while maintaining exceptional individual contributor velocity (370+ merged PRs) on frameworks handling 10,000+ hourly transactions. Led a zero-downtime search infrastructure overhaul migrating from Elasticsearch to OpenSearch, and elevated core system test coverage from 55% to 80% via AI-assisted edge case discovery.
• 3Pillar Global (Lead Software Engineer): Modernized legacy enterprise monolithic codebases by enforcing structured Domain-Driven Design (DDD) principles and Service Objects, and designed a comprehensive end-to-end multi-region i18n framework from scratch.
• MagmaLabs (Senior Engineering Manager): Guided company-wide high-throughput e-commerce integrations, scaling progressive checkout workflows, advanced subscription layers, and complex multi-region payment gateways across the Spree and Solidus ecosystems.

I help engineering teams adopt AI tooling across Ruby, Elixir, and Rust through consulting and open-source tools.

• 💼 Consulting: AI context audits, MCP server implementation, skill pack customization, and eval-driven quality improvement.
• 💬 Open source: Discuss rails-ai-bridge, ruby-skill-bench, brigid, or elixir-phoenix-skills via GitHub issues or discussions.
• 📧 Contact: LinkedIn or ismael.marin@gmail.com