JRM
ES
Portfolio — Information Systems Engineer

Juan IgnacioRodríguezMariani

Scale 1:1UNS '26

Information Systems Engineer · Full-stack & AI-augmented development

shipping Cuonti — 50+ active clients in production

About

I build and operate production web platforms end to end: from API design to deployment and release management on Linux infrastructure I run myself. Information Systems Engineer, graduated from Universidad Nacional del Sur.

By day I work as a Software Engineer at NexoSmart (Laravel + Next.js). Alongside that, I co-created Cuonti, a multi-tenant tax-automation SaaS used in production by 50+ active clients. Agentic AI tooling — Claude Code in particular — is a core part of how I ship.

Information Systems EngineeringUniversidad Nacional del Sur (2026)

Juan Ignacio Rodríguez Mariani
Fig. 01 — Bahía Blanca, Argentina

Experience

NexoSmart

Software Engineer

Sep 2025 – present · Software Developer Intern Jun–Sep 2025

  • Develop REST APIs in Laravel and frontend features in Next.js for client-facing products.
  • Reduced response times on the highest-traffic endpoints with database indexes and Laravel query caching.
  • Own server configuration, production deployments and release cycles on Linux with Nginx.

XenovaIT

Lead Engineer & Technical Director · self-employed

Jan 2025 – present

  • Independent software practice run in parallel with my role at NexoSmart — client engagements handled end to end: requirements scoping, architecture and operations.
  • Built and operate a private management platform used by 10+ partner companies in the food industry, integrating third-party services into a single Laravel + Next.js system.
  • Built a management platform for 5taPata (retail) that replaced manual administration and inventory workflows with one web application.

Le Pas Sage

Freelance Full-stack Developer

2024 – 2025

  • Built an e-commerce platform end to end: Laravel backend, Next.js storefront and payment integration.

Projects

3.1In production

Cuonti

Co-creator & full-stack engineer

Multi-tenant SaaS that automates Argentine tax compliance for accounting firms. In production with 50+ active clients and 150+ registered accounts.

50+
active clients
150+
registered accounts
  • Native integrations with ARCA (Argentina's federal tax agency): electronic invoicing and F.931 payroll-tax filing generation.
  • In-product AI assistant that lets accounting firms query and operate on their tax data.
  • Multi-tenant architecture serving every firm from a single deployment.
  • Laravel
  • React
  • MySQL
3.2Public repo

Agro BCS

Engineer — UNS & INTA research collaboration

Mobile app built with INTA researchers that estimates bovine body condition scores (BCS) from field photos, using image processing to replace manual visual scoring.

  • Thesis project (with Ian Sebalt and Lucas Bazán), directed by Dr. Dana Urribarri and Dr. Martín Larrea, in collaboration with INTA.
  • Already in use on real farms in the region — works offline in the field.
  • Covered by regional press and industry media.
  • TypeScript
  • Python
3.3Self-hosted

Production VPS

Owner & operator

Self-managed Ubuntu VPS where my applications run in production: Nginx reverse proxy, Cloudflare in front, and deploy pipelines I built and operate myself.

  • Full ownership of the path from merge to production: server config, TLS, releases and rollbacks.
  • Ubuntu
  • Nginx
  • Cloudflare
  • CI/CD

How I work

Not autocomplete — agentic workflows. I run AI agents as part of my engineering process on real production codebases, with the same rigor as any other code: reviewed, tested and verified before it ships.

4.1

Multi-agent workflows

Planning agents, parallel codebase explorers and adversarial reviewers orchestrated on real features — not toy demos.

4.2

Custom skills & commands

Reusable skills and slash commands that encode repeatable engineering tasks: design systems, code review checklists, deploy routines.

4.3

MCP servers in daily work

Model Context Protocol servers wired into the workflow — project management, documentation and design tools reachable from the agent session.

4.4

Verification before done

Every AI-assisted change runs the same gauntlet: builds, tests, lint and a human review. AI speeds up the loop; it doesn't skip it.

Claude Code

daily driver — agents, skills, MCP, multi-agent orchestration

Codex

secondary agent for parallel work

Gemini

API integrations in shipped products

Skills

Backend

daily at NexoSmart and in Cuonti

  • Laravel / PHP
  • Node.js
  • REST API design
  • MySQL
  • PostgreSQL
  • Redis

Frontend

production UIs for client-facing products

  • React
  • Next.js
  • TypeScript
  • Tailwind CSS

Infra & DevOps

self-managed production VPS

  • Linux (Ubuntu)
  • Nginx
  • Cloudflare
  • CI/CD
  • Git

AI tooling

agentic workflows as part of how I ship

  • Claude Code
  • Multi-agent workflows
  • MCP servers
  • Codex
  • Gemini

Languages

  • Spanish (native)
  • English (B2)
  • Italian (basic)

Contact

Let's talk

Open to interesting engineering work. The fastest way to reach me is email — I answer.

juanirmariani@gmail.com