Before building anything I ask what real value it gives the user, and whether it comes from a hypothesis of ours or from something the user needs. Today I build FIJI at Prizmstack, a platform that prices a real estate business from how it has been operating.
I came to product at Banana Software: four years, from trainee to product engineer, projects run end to end, B2C and B2B platforms with payment methods integrated. That is where I realised I had been asking that question by default, and that my whole job revolved around it. Clients arrived with lists of features that did not follow that line, and I negotiated with data. The answer was almost always to do less, because what was planned did not earn what it cost, and doing less is what lets you iterate fast. Sprint predictability went from 60% to 90% once I stopped estimating by eye and modelled real effort.
Since 2026 I pick the questions myself. I write about health, which comes from my bioengineering background and a machine learning thesis on real clinical data, and about insurance, a market that publishes everything and explains little. Both are businesses where the company decides for you what you are covered for, how much you are charged and what you get shown before you buy, and that decision leaves a public trace: every payment to a doctor that the pharmaceutical industry has to disclose, the balance sheets insurers file with the regulator, workers' compensation statistics, what they show (or don't) on their own digital channels.
In each case I rebuild the problem and the players involved, and take it through to a decision: how much money is at stake, what the first step is, which metrics measure it by segment, and what is not worth building.
If your team has a roadmap full of features nobody has proved are worth what they cost, a decision that keeps getting made by eye because nobody has worked out the number, or a launch you cannot tell whether it worked, that is what I solve. Email me with yours and I'll tell you what I would do. And if it's a product you're thinking about and you want a second head on it, book a call.
16 case studies published since June 2026, one every 5 days.
See all case studies
I build FIJI, a valuation platform for real estate businesses. I lead the migration from no-code (Bubble) to Next.js and build the simulation engine that turns operating metrics into financial valuations.
I teach Digital Techniques I to 4th-year Electronics students: Boolean algebra, Karnaugh maps, combinational and sequential logic. I also help plan the academic program.
My longest experience, and where I learned to build product. I came in as a Trainee and left as a Product Engineer, four years later. I wrote about how that path went and what the role does in practice.
As Product Engineer (Aug 2025 – Jan 2026), I improved sprint predictability from 60% to 90% by modeling real effort. Removed ~20% of low-value features by prioritizing by impact.
As Semi Senior (Jul 2024 – Jul 2025), I led 4 end-to-end projects with teams of 3-4 people. B2C and B2B platforms with up to 1,500 active users. Integrated payments with Mercado Pago, PayPal and Stripe; optimized B2C checkout, increasing booking completion rates.
Fintech product 0→1 to simplify tax management for freelancers in Argentina. Participated in the Techstars W24 batch (Stellar & MoneyGram Accelerator) in New York as part of the founding team.
My first full-time role, before moving into software: I planned preventive maintenance at a national level on Microsoft Dynamics NAV. It lasted three months and showed me how a large company designs its processes and measures its performance.
Biomedical Engineering
Buenos Aires
Thesis: 2-year mortality prediction and KRAS gene mutation in colorectal cancer liver metastasis, developed in collaboration with Hospital Italiano de Buenos Aires. The research came from a need the hospital had: predicting how an oncology patient would evolve from their historical data.
What I delivered was an algorithm that takes a new patient and estimates their risk, measured for precision and accuracy on data the model had not seen. I built it with radiomics and machine learning on real clinical data.
I was also a teaching assistant in Biomedical Image Processing and in Biosensors.
Electronics Technician
Posadas, Argentina
Technical training in analog electronics, digital electronics and communications. I teach there now.
Spanish · Native
English · Intermediate (B1)
Duolingo English Test · 85/160 · May 2026I do the analysis in Python and SQL, and the measurement with PostHog. When I need to see the product working, I prototype it in Next.js and TypeScript. I've integrated payments with Stripe, Mercado Pago and PayPal, and moved a product off Bubble to rewrite it in Next.js.
If you want to talk, reach me via . I reply within 24h.