Intro / Name
Applied AI engineer / Product engineer, Buenos Aires
I build applied AI systems that keep model output reviewable: structured extraction, voice workflows, tool and API orchestration, deterministic controls, evaluation, failure recovery, and explicit human approval.
At Acelera, I lead applied AI engineering across workflow discovery, orchestration, and tested pilot handoff. At Rely, I lead AI and product engineering for a live operating platform used across 100+ US LLC formations and 6+ connected services.
Work / Product
what I build
Selected systems across applied AI, operational software, and physical product delivery. The model, controls, tests, and evidence stay explicit.
AI + company operations / co-founder
AI-assisted company operations behind 100+ US LLC formations and approximately USD 50K in formation packages.
Mapped filings, banking, payments, documents, deadlines, and support into one persistent company context across 6+ connected services.
AI assembles context, identifies missing information, and recommends next actions; filings, payments, and other external writes stay behind explicit approval.
Offline document AI / 139 tests
A local Python pipeline that turns sensitive Argentine legal PDFs into structured, reviewable JSON without uploading them to a SaaS model.
Qwen2.5-7B runs through llama.cpp, so legal PDFs and model inference do not leave the local environment.
A field-level golden set checks extraction quality; schema validation, JSON repair, source anchoring, and retries turn model output into a reviewable workflow.
Errors stay visible and tamper-evident audit logs preserve what ran, what changed, and what still requires human review.
Applied AI engineering / co-founder
I turn operational workflows into bounded AI systems with tool calls, deterministic rules, human approval, evaluation, and explicit failure recovery.
The LLM interprets natural-language requests; deterministic rules control slot availability, required documents, consent, and which actions the system may take.
Provider failure, missing documents, unavailable slots, and clinical-risk signals route to explicit fallbacks or human review instead of an invented answer.
The prototype is covered by 95 automated tests and is presented as a pilot artifact, not as a production deployment.
Paisanos NGO hackathon / winner
Four of us interviewed NGOs and donors, built a round-up donation concept, and won the event.
We spoke with NGO teams and donors before choosing a purchase round-up as the recurring-donation mechanic.
We built the prototype during the event and won the Paisanos NGO hackathon.
Physical product / 150 units
Lemon asked if a generated product image could become real. Joaquín and I turned the prompt into 150 working market displays.
Joaquín modeled the enclosure. I handled firmware, hardware integration, printing, assembly, and the three 3D printers used for production.
We iterated on the ESP32-S3 build and made 150 physical units for Lemon.
Side / Projects
side projects
Onchain products I build outside my core roles, where system state, custody, and verification have to remain inspectable.
Side project / Solana + Telegram
A live side project combining Telegram games, wallet flows, Solana contracts, and a public result verifier.
Connected wallet actions, game UX, contract state, oracle reveal, and public verification into one product path.
Timba is a side project. The contracts are public; bot and verifier implementation evidence is shown through sanitized product surfaces.
Side project / market operations
A live multi-user operating system for discovering sports markets, routing strategy decisions, executing through guarded wallets, and reconciling order state against on-chain activity.
Match live sports data with Gamma metadata and CLOB liquidity before an opportunity reaches a strategy.
Route each decision through the correct user, strategy, mode, wallet, balance check, and pre-trade guard.
Track orders through fills, active exits, reconciliation, dashboard state, and Telegram diagnostics.
Range / Background
background
Contact / Line
get in touch
Looking for Applied AI and Product Engineering roles where I can own the workflow from model behavior and evaluation through backend, frontend, integrations, and production operations. Remote preferred; also available for hybrid or on-site work.