name:     Gabriel Cruz Ferreira
role:     Salesforce Architect & Full-Cycle Specialist
based_in: Porto Alegre, RS — Brasil 🇧🇷
focus:
  - 💼  Salesforce Engineering Lead at Race Communication — a California-based telecom 🇺🇸
  - 🏛️  Architect @ XCL-Consulting
  - 🤖  Building autonomous AI squads to ship products end-to-end
  - 📖  Mentoring interns and growing the next wave of Salesforce talent
  - ⚙️  Designing and maintaining Salesforce CI/CD pipelines
ask_me_about:
  - Salesforce — development, administration, consulting, architecture

My focus right now extends well beyond the Salesforce platform. I'm investing heavily in agentic AI engineering — designing autonomous, multi-agent squads that can take ownership of an entire delivery cycle, from a fuzzy idea on a Jira ticket all the way to a merge request waiting for human review.

In practice, this looks like several connected workstreams:

• 🎫 Jira task automation — AI agents that watch incoming tickets, parse the requirements, pull the right context from the codebase, documentation and previous issues, and then triage, refine, estimate and decompose them before any human touches them. The goal is to remove most of the grooming overhead and start every ticket with a clear, actionable plan.

• 🛠️ AI-powered CI/CD pipelines — embedding model-driven steps directly into Jenkins, GitLab and Azure DevOps flows so that test generation, static analysis, code review, security scanning, release notes and deployment validation happen automatically as part of the build. The pipeline becomes an active reviewer instead of a passive runner.

• 🤝 End-to-end autonomous delivery — the bigger vision: an agent picks up a Jira issue, plans the implementation, writes the code, runs and iterates on the tests, validates against acceptance criteria, and opens a merge request in the Git client with a clean diff and a written summary, ready for a human to approve. Effectively, an AI teammate that follows the same workflow your engineers do.

• 🧠 Engineering rigor for prompts — versioning prompts, building evaluation harnesses, defining guardrails and observability for agent behavior, and treating the whole orchestration layer as production software. The same discipline I apply to metadata, Apex and Salesforce DevOps, now applied to LLM-driven workflows.

The bigger bet is that shipping software is becoming as much about supervising agents as it is about writing code yourself — and Salesforce delivery, with its strong metadata model and well-defined CI/CD patterns, is a fantastic place to prove it.

☁️ Salesforce Ecosystem

🔄 DevOps & CI/CD

💻 Languages & Web

🤖 AI & Automation