Skip to content
View felipefontoura's full-sized avatar
🎯
Focusing
🎯
Focusing
503 followers · 128 following

Achievements

Achievement: Pair Extraordinairex2Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLOAchievement: Public SponsorAchievement: Arctic Code Vault ContributorAchievement: Starstruck

Achievements

Achievement: Pair Extraordinairex2Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLOAchievement: Public SponsorAchievement: Arctic Code Vault ContributorAchievement: Starstruck

Block or report felipefontoura

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
felipefontoura/README.md

Felipe Fontoura

Engineered with AI. Production systems that can't fail.

Staff Engineer · Production Systems with AI · builder

I build secure, reliable, scalable systems with AI agents. 25+ years shipping production software, from satellite kernel drivers at INPE to the global checkout serving 20M+ customers at TUI Group (€23B revenue), to a crypto fintech built solo in 70 days.

→ felipefontoura.com

What I built

A crypto fintech on Bitcoin's Liquid Network: a payment gateway, an OTC exchange, and on-chain settlement. Engineered solo in 70 days with AI agents. Private repo.

TypeScript Node.js Fastify PostgreSQL Kubernetes Terraform Redis BullMQ

  • 13-app monorepo, 3 Fastify APIs, 3 isolated PostgreSQL databases
  • PIX gateway, 4 PSPs behind a factory pattern (HMAC-SHA256, mTLS to BACEN)
  • OTC exchange engine: rolling 24h VWAP, 5 sources, asymmetric spreads
  • On-chain settler: confirmation polling, deviation refunds, crash recovery
  • OAuth 2.1 / OIDC: 5 roles, 19 scopes, JWKS, M2M, TOTP 2FA
  • Kubernetes (DOKS, HPA 2-10), Prometheus and Grafana, CI/CD

→ Read the case study

How I build

Build the right thing, engineer it to hold, direct the agents that write it. The bar has always been production: satellites where a failed signal has no patch, a €23B checkout at 99.9% uptime, a fintech moving real money. The AI writes the code. The architecture, the specs, and the call on what "done" means are mine.

The method is plain: structured specs (requirements, design, tasks) are the source of truth, agents execute against them, humans review and ship. I have trained 400+ people in it.

Open source

  • pi-sdd-kit: spec-driven development as skills for the Pi coding agent. Steering docs as durable memory, .status approval gates, EARS, and a gated PRD to review pipeline.
  • pi-skill-model-handoff: let each Pi skill pick its own model and thinking level from the SKILL.md frontmatter.
  • AI engineering dotfiles: slash commands, agent teams, review pipeline.

Writing

Notes on building production software with AI coding agents: harness engineering, spec-driven development, context, and architecture.

→ felipefontoura.com/articles

Stack

Backend Infrastructure
Node.js / TypeScript Kubernetes (DOKS)
Fastify / NestJS Terraform
Drizzle ORM / Prisma Docker
PostgreSQL GitHub Actions CI/CD
BullMQ / Redis Prometheus / Grafana

Links

Popular repositories Loading

  1. doc2md doc2md Public

    Convert documents to Markdown format through a simple API service.

    Python 29 20

  2. ubinkaze ubinkaze Public

    An comprehensive post-installation configuration tool for Ubuntu servers.

    Shell 19 18

  3. bento bento Public

    Bento turns a fresh Ubuntu/Debian VPS into a hardened Docker Swarm server with Traefik, Portainer, TLS, and deployable apps through a guided terminal menu.

    Shell 17 20

  4. deepseek-local deepseek-local Public

    My DeepSeek-R1 local machine test.

    Python 14 13

  5. pi-sdd-kit pi-sdd-kit Public

    Spec-Driven Development skill pack for the Pi coding agent: steering docs, .status gates, EARS, and a PRD→SPEC→TASKS→EXEC→REVIEW pipeline.

    11 1

  6. pyzerox-demo pyzerox-demo Public

    A demonstration project showcasing the usage of PyZerox library for processing PDF documents using GPT-4 models.

    Python 8 4