Engineering intelligent products
I design and ship voice agents
from first commit to production.
Based in San Ignacio, Paraguay
--:--:-- · GMT−3
Working remotely, worldwide
I'maseniorfull-stackengineerwhohasbeenshippingLLMstoproductionsince2022.IturnambitiousAIideasintoproductspeopleactuallyuse—voiceagents,RAGpipelines,autonomousworkflows—andownthemend-to-end,fromarchitecturetoadoption.
Five-plus years across international startups in the US, Argentina, Mexico and Paraguay. I've been the first engineer more than once — the person who turns a founder's whiteboard into a system real users depend on.
Today I combine product strategy with hands-on execution as Entrepreneur in Residence at InzaiQ, validating AI-native products from discovery to delivery.
Years shipping production software
Shipping LLMs to production since
Startup teams across 4 countries
Students taught to code
From a voice AI that interviews your grandparents to Git-backed memory for agents — a few products and experiments I'm proud of.
“Tell me about the house you grew up in…”
An AI voice interviewer that calls you and preserves your life story. Real-time synthesis, telephony and transcription — LLMs in production from day one.
Verifiable agent memory backed by Git. Mandu'a means “to remember” in Guarani — decisions, corrections and evidence live in native Git history.
A long-term memory app for capturing, searching and retrieving personal knowledge with semantic search.
Generate Spotify playlists from a prompt — authenticated, deployed and ready to press play.
Finds YC companies that match your profile and the kind of company you want to join, using embedding similarity.
Best value → model · β
Recommends the quality-to-price optimal AI model for a task, driven by Artificial Analysis benchmark snapshots.
A rare overlap: deep applied-AI experience, full-stack craft, and the product instincts of someone who has been an early engineer again and again.
LLM features that survive real users.
Conversational bots, retrieval pipelines and agents wired into real workflows — in production since 2022.
Systems that quietly do the work.
Agent graphs, messaging bots and webhook pipelines that connect AI to the tools teams already use.
From schema to pixel, end-to-end.
Typed, tested, cloud-deployed products — frontend, backend, data and infra owned by one pair of hands.
Founder-minded execution.
Discovery, scoping and architecture with non-technical stakeholders — then rapid MVPs that iterate on feedback.