Jose Nobile — AI Engineering Profile

I build payment systems, decompose monoliths, harden security, and deploy production AI systems. Full-stack platform engineer owning outcomes end-to-end across 29 microservices, with the OpenClaw AI gateway (Claude Opus 4.6 + GPT-5.4 failover) running 24/7 on Docker.

Solutions I’ve Built

Real problems solved in production. Each card: what was broken, what I built, and the measurable impact.

PAYMENTS

Multi-Country Payment Infrastructure

Problem: No unified payment flow, manual onboarding, missing fraud prevention, no tax compliance across Latin America.

  • Built complete Stripe Connect Express integration: onboarding, fee management, connected accounts
  • Stripe Checkout migration with 3DS enforcement
  • No-show fee system spanning 10+ repos (Check-In, Android, iOS, Web, Payments, Lessons, Reports)
  • Tax compliance: Chile (SII/Boleta Exenta), Peru, Colombia (Siigo electronic invoicing)
  • Fraud prevention: stolen card detection + payment blocking
  • Production incident response: payout rescue, 3DS failures affecting multiple studios

Impact: End-to-end payment infrastructure serving multiple countries with automated tax compliance and fraud prevention.

SECURITY

Platform Security Hardening

Problem: No rate limiting, SQL injection vectors, hardcoded credentials, no session-based auth.

  • Redis-backed sliding-window rate limiting across 29 microservices (Lua scripts, cluster-safe)
  • Cloudflare edge rate-limit blocker with expiry sweeper
  • SQL injection remediation (parameterized queries)
  • Removed hardcoded credentials from codebases
  • Built session-based dynamic authentication with Authorizer service (from scratch)
  • Geolocation-based login authorization

Impact: Zero successful attacks after hardening. Defense-in-depth across edge, application, and database layers.

ARCHITECTURE

Monolith → 26 Microservices

Problem: Monolithic backend with tight coupling, single point of failure, unmanageable deployments.

  • Decomposed a monolithic backend into 26 independently deployable services
  • Each service: Dockerfile, Helm chart, GitLab CI pipeline, health endpoints, K8s resource tuning
  • Queue architecture with ZeroMQ for async processing
  • MySQL connection resilience across all services (auto-reconnect)
  • Fixed connection pool exhaustion causing OOM crashes in production

Impact: Independent deploys, isolated failures, horizontal scaling per service. Zero-downtime releases.

PLATFORM

Webhook Delivery Engine

Problem: No way for third-party integrations to receive real-time platform events.

  • Built from scratch: idempotent delivery, exponential retries, HMAC signing, multi-config fan-out
  • Management API + Admin UI with event log and developer docs
  • 10+ event types across the entire platform

Impact: Enabled third-party integrations with reliable, secure, auditable event delivery.

DEVOPS

CI/CD & Infrastructure

Problem: Manual deployments, no container orchestration, slow builds, OOM failures in production.

  • Kubernetes on GKE with Istio service mesh
  • Docker to Kaniko migration across 10+ repos
  • Helm charts for every microservice
  • White-label app automation: 41 branded Android apps from JSON config
  • OOM prevention (memory tuning), GeoLite2 caching (fixed 504s)
  • GitLab Runner (concurrent: 10)

Impact: Fully automated pipeline from commit to production. 41 branded apps built from a single config.

FULL-STACK

Full-Stack Delivery

Problem: Slow admin panel, fragmented frontend stack, no email analytics, missing mobile features.

  • Angular admin panel: lazy loading overhaul, 730 KB savings, Lighthouse 90+
  • React + Vue.js consumer apps with Stripe Checkout integration
  • Android + iOS native: no-show fees, coupon support, biometric check-in
  • Email infrastructure: send/delivery/bounce analytics, 10 min to 10 sec reporting

Impact: Admin panel 730 KB lighter. Email reporting 60x faster. Native apps with complete feature parity.

Accomplishment Timeline

30 high-impact accomplishments across 3+ years. Each card represents a meaningful production milestone.

30 milestones • Dec 2022 – Apr 2026 • Auto-scrolling

AI-Augmented Engineering

AI is not a buzzword in my workflow. It is a force multiplier I use daily to ship faster and with higher quality.

PARALLEL OPS

8 Concurrent Claude Code Instances

Ran overnight for tax billing verification across 3 countries, 7 repos, 1,740 tests, 11 commits. One engineer, eight parallel AI agents.

CUSTOM SKILLS

8 Custom Claude Code Skills

/testing, /deploying, /debugging, /developing, /mr-review, /jira-docs, /planning, /cli-tools. Each skill encapsulates domain-specific workflows for the codebase.

SCALE

2,205 Claude Code Sessions

Across 26 active projects. Not just chat prompts: structured agent sessions for code generation, review, testing, and deployment. 480 commits in the last 30 days alone.

AI AGENT

OpenClaw AI Gateway

Docker container with Claude Opus 4.6 + GPT-5.4 failover. Multi-channel AI assistant (WhatsApp + Telegram) with voice transcription, vector memory (sqlite-vec), and billing proxy routing through subscription.

OpenClaw Deployment Portfolio

A real production AI system I built and operate 24/7. Not a demo. Not a prototype. Running on my hardware right now.

GATEWAY

AI Gateway & Model Failover

  • Docker container with Claude Opus 4.6 primary + GPT-5.4 automatic failover
  • Billing proxy routes through subscription instead of API credits
  • Model resilience: automatic fallback on rate limits, errors, or outages
  • Vector memory with local embeddings (sqlite-vec) for persistent context
CHANNELS

Multi-Channel Messaging

  • WhatsApp + Telegram channels with 24/7 availability
  • Local Whisper STT for voice transcription (CUDA-accelerated)
  • LuxTTS cloned voice synthesis (not cloud APIs)
  • Self-hosted Telegram Bot API for large file handling
BROWSER

Remote Browser Control

  • Chrome DevTools Protocol (CDP) from WSL2/Docker
  • AI-driven web navigation, form filling, screenshot capture
  • Puppeteer automation for banking, scraping, monitoring
IOT

IoT & Health Automation

  • Xiaomi Scale BLE health tracking via systemd daemon
  • Auto-deploy dashboard with body composition data
  • 375+ measurements, 30+ metrics, 14+ months continuous
AUTOMATION

Cron Jobs & Services

  • 9 automated cron jobs (watchdog, health deploy, currency rates, backups, etc.)
  • 6 custom systemd services running 24/7
  • Bank monitoring: headless Puppeteer scraping every 15 min
  • Self-healing watchdogs with automatic restart on failure
ARCHITECTURE

Infrastructure Stack

  • Docker Compose orchestration (rootless)
  • 10,159 source files across the OpenClaw workspace
  • GitLab Runner for CI/CD (user-mode)
  • RTX 5090 GPU for local AI inference (Whisper, TTS, Vision)

Autonomous Systems I Built

Not scripts. Autonomous systems with watchdogs, failure recovery, and continuous operation.

AUTONOMOUS

Xiaomi Scale BLE Body Composition Tracker

  • BLE health tracking via xiaomi-scale-daemon systemd service
  • ADB + uiautomator scrapes Xiaomi Home app automatically
  • 375 measurements over 14 months, 30+ body metrics tracked
  • Auto-syncs to Google Sheets + interactive web dashboard
  • Watchdog + health alerts with Windows sound notifications

Why it matters: IoT integration, reverse engineering, daemon reliability.

AUTONOMOUS

Banking Monitor

  • Puppeteer Extra with Stealth Plugin bypasses Imperva WAF
  • Local vision AI (Qwen2.5-VL-7B) solves hCaptcha autonomously
  • Headless Puppeteer scraping every 15 min from Colombian corporate banking
  • Cookie persistence, failure recovery, watchdog

Why it matters: browser automation at production scale with AI-powered CAPTCHA solving.

AI AGENT

OpenClaw AI Gateway

  • Docker container: Claude Opus 4.6 primary + GPT-5.4 automatic failover
  • WhatsApp + Telegram with local Whisper STT + cloned voice TTS
  • Remote browser control via Chrome CDP from WSL2/Docker
  • Vector memory with local embeddings (sqlite-vec), billing proxy routing
  • 9 cron jobs, 6 systemd services, self-healing watchdogs

Why it matters: full production AI gateway with model failover, local inference, and autonomous operation.

Technical Depth

AI & Agents
OpenClaw Ecosystem
Backend
Node.js TypeScript Python NestJS Fastify
Frontend
Mobile
DevOps
Cloud
Data & Integrations
AI / ML
Whisper Qwen3-TTS LoRA fine-tuning Video generation sqlite-vec Chrome CDP

Hardware & AI Investment

MSI Raider 18 HX (2025)

  • RTX 5090 — 24 GB GDDR7
  • 64 GB DDR5-6400
  • 6 TB NVMe (PCIe 5.0 + PCIe 4.0)

AI Subscriptions

  • $257/month (Claude Max, ChatGPT, Gemini, Perplexity)
  • 6 local AI models on RTX 5090
  • 6 custom systemd services always running

Total Monthly AI Investment

~$526
Hardware amortization + subscriptions + compute

Published Knowledge

Consulting & Training

Remote sessions via video call. From Claude Code fundamentals to production AI agent architecture.

Single Hour
$95.90
USD / hour
No commitment
5-Hour Pack
$81.50
USD / hour
-15%
$407.50 USD
20-Hour Pack
$47.95
USD / hour
-50%
$959.00 USD

Payment: Binance (USDT) • PayPal • Credit/Debit card link • Prepaid packs