Product & Software Engineer

I own products end to end: the roadmap, the code, the infrastructure, and the AI inside. I also lead a team of interns and build each of them a real work experience.

Let’s talk

Currently: Product & Engineering Lead at(opens in a new tab)

Founder of(opens in a new tab)

work

Production systems, built under real constraints.

AI-assisted help desk for a managed-service provider

2026 · Solo

A managed-service provider with a 10-person support team handling ~250 tickets/month.

A human-in-the-loop AI copilot. It classifies priority, queries recent context, and prepares draft replies. Nothing closes without human approval.

It runs on an open-weight model through NVIDIA's hosted API. A 28 rpm rate governor queues calls by priority, so an agent waiting on a draft is never starved, and a circuit breaker falls back to deterministic rules when the provider fails.

Agents approve drafts directly in their workflow, and the repetitive triage load was handled without hiring another L1.

Stack

Next.js
FastAPI
Postgres
Redis
SuperTokens
Docker
Railway
MongoDB
Langfuse
NVIDIA NIM
GLM 5.3

Fleet monitoring for a managed-service provider

2026 · Solo

Infrastructure monitor watching client machines across a tree of clients, sites, and agents. Active in production.

An agent on each machine pulls work over WebSockets, backed by a Go server, Postgres, and a React dashboard.

Mutual TLS with per-agent certificates prevents reverse-proxy stripping. Rolling 48-hour certificate renewals make revocation enforceable.

Full remote command execution, an interactive xterm.js terminal, multi-OS patch management, and Linux CVE scanning.

Stack

Go
React
Postgres
mTLS
WebSocket
xterm.js
Docker

Additional client work is under NDA. Architecture details and operational metrics available on request.

Let’s talk

side projects

Built on my own time.

Own product · Waitlist open · 2026

Jobs in Chubut, with the salary in plain sight.

A job platform for Chubut, in Argentine Patagonia. Every listing has to show its salary range, and a reskilling engine points workers from declining sectors toward wind farms and shipbuilding. I founded it and built the product.

Visit aceler.ar

Mantle

2026

Keeps sensitive data out of the prompt.

A Manifest V3 extension that freezes the send on ChatGPT, Claude, and Gemini until it has read the prompt. Cards, API keys, ID numbers, and addresses are swapped for tokens held in an ephemeral session vault, then restored in the reply, so the model never receives the real values and the user never sees the tokens. Detection runs in an offscreen document so scanning cannot block the page.

TypeScript
Manifest V3
Bun
Web Crypto

Source

experience

From the support desk to owning the product.

Product & Engineering Lead

part of LA Tecnologías

Ticketera grew into a team, so I went from building alone to leading: product, infrastructure, and the interns who grow with it.

Own the roadmap and the product decisions. I work directly with the C-level and sit down with marketing and growth.
Still build across the stack and run the infrastructure: Railway, deploys, and domains.
Lead a team of interns. I help each one find what they are good at and shape the work around it: one is growing into a product engineer, another toward design engineering, each at their own pace.

Next.js
FastAPI
Postgres
Docker
Railway

Aug 2026 to PresentRemote · Argentina

AI Engineer

Worked solo as an individual contributor. Owned Ticketera, a multi-tenant help desk SaaS for MSPs, from research and architecture through to containerized production.

Put an LLM (GLM 5.3 through NVIDIA NIM's hosted API) in the ticket pipeline to classify priority and suggest replies. A tiered rate governor, a circuit breaker, and deterministic fallbacks keep it working when the provider is limited or down.
Wired an eval gate into CI for the AI features: ticket classification, report insights, and rule generation.
Designed the multi-tenant PostgreSQL schema and the deploy stack (Railway · Docker · Nginx). Owned the roadmap, API contracts, and CI/CD.
Built the delivery workflow around agentic tooling: Claude Code custom skills, MCP servers, and sub-agents.

Python
FastAPI
React
Postgres
Docker
Railway
GLM 5.3

Mar 2026 to Aug 2026Remote · Argentina

Junior Software Engineer

Started as an L2 support intern, resolving production network issues on MikroTik and automating routine maintenance with Bash and Python. Then moved into engineering: built internal Python CLI tools that cut incident resolution time 30%, built the Linux infrastructure (Debian VMs), and standardized Docker deployments.

Aug 2025 to Mar 2026Remote · Argentina

Education

B.S. Computer Science (Licenciatura en Informática)
Universidad Empresarial Siglo 21
Analyst in Information Systems (Analista en Sistemas)
Universidad Empresarial Siglo 21

Stack

TypeScript
Python
Go
Next.js
FastAPI
Postgres
Redis
Docker
Railway
Vercel

LLM agents · RAG with pgvector · Embeddings · Prompt engineering · Evals & guardrails · MCP / tool-use

contact

Let's talk about what you want to ship.