Skip to content
View jrogatis's full-sized avatar
🤔
solving problems...
🤔
solving problems...

Highlights

  • Pro

Organizations

@rogatisApps

Block or report jrogatis

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
jrogatis/README.md

JP — Jean Philip de Rogatis

Engineering executive · AI-native operator · Father · Skydiver

I build and lead engineering organizations — and right now I'm obsessed with what AI is doing to how we ship software. Not the hype. The operating model.

~30 years across fintech, cloud, and telecom. Most recently leading 100+ engineers shipping platforms that process millions of transactions/day in regulated environments. Based between São Paulo and Miami.


🤖 What I've shipped with AI (last 18 months)

  • AI-augmented engineering transformation across a 100-person org — DXI 67, 86% of devs saving 2+ hrs/week, ~20% velocity gain measured end-to-end (not vibes — instrumented)
  • Built internal agents & tooling (Node.js + TypeScript) for capacity analysis, deploy-risk scoring, and engineering metrics — moved decisions from gut to data
  • Platform strategy & governance for Claude vs Gemini Enterprise adoption in a BACEN-regulated fintech — dual-track rollout, InfoSec alignment, cost model
  • Live in Claude Code, MCP servers, and agentic workflows as my daily operating layer — not as side experiments

I treat AI tooling the way I treat infra: measured, idempotent, observable. If it doesn't move DXI or lead time, it's a demo — not a system.


🛠️ Stack

AI-native engineering — Claude (Code, Desktop, Enterprise), Copilot, MCP, agentic workflows, prompt engineering, eval loops

Cloud & Architecture — AWS Serverless (Lambda, EventBridge, Step Functions, DynamoDB) · event-sourcing · idempotency · stateless systems

Languages — Node.js · TypeScript · .NET · JavaScript

Domains — Payments · BaaS · Ledgers · Regulated fintech (BACEN / PIX)


🚀 Currently

  • 🤖 Shipping personal agents and internal tooling on Node.js + AWS Serverless
  • 🧪 Running my own evals on Claude, Gemini, and Copilot for real engineering workflows
  • 🛰️ Building with MCP — connecting AI to where the work actually lives
  • 🪂 Skydiving when calendars allow

💡 How I think about engineering in the AI era

  • AI doesn't replace engineers. It exposes the org. If devs aren't shipping faster with it, the bottleneck was never typing speed.
  • Done = value in production. AI moved the bar — speed without observability and idempotency is just faster regret.
  • Optimize the queues, not the keystrokes. Most lead time is waiting, not building. AI helps less than process does.
  • Measure DXI, not vibes. Engineering intelligence > happiness surveys.
  • Simple beats clever. Boring beats fragile.

🤝 Connect

LinkedIn · YouTube

Pinned Loading

  1. meeting-for-good meeting-for-good Public

    Forked from freeCodeCamp/meeting-for-good

    A meeting coordination app for your team

    JavaScript 1 1