Thoughts on AI product management, quantitative finance, and building at scale.

The 28-Second Bounce

A new user signed up, opened the app, and left 28 seconds later without adding a single contact. Here's why I instrumented the funnel before touching the redesign — and what I shipped.

Context Engineering Is the Job Now

Running a fleet of AI agents only works if the context is ordered. My setup — one fact per memory file, an index that loads every session, and chats scoped by topic — is what keeps parallel agentic work fast instead of chaotic. The payoff compounds.

Choosing Not to Scale

The most underrated engineering decision is removing the reason you needed to scale in the first place, instead of adding infrastructure or spend. A note on solving by subtraction.

Running AI Agents Like a Team

Running multiple AI coding agents in parallel, each owning a slice, while the human reviews and integrates. The bottleneck moves from building speed to clarity of specification and judgment.

When to Build a Focused Product on Top of a Generic AI Agent

Generic AI agents can do almost anything. That is exactly why a focused, opinionated product can still win the specific job. A product-strategy note, with Scout as the example.

The WASP Framework: From Challenge to AI MVP in One Week

A hypothesis-driven methodology for building AI products at enterprise speed. Three phases: Challenge Scoping, Problem Space, Solution Space.

Zero Integration Bugs: How I Built an API Contract Platform for My Teams

Auto-generated test cases from API schemas. Frontend and backend agree on contracts from day zero.

Building a B2B SaaS from Scratch: The Pepino Brand Story

Co-founded a Digital Asset Management SaaS. Closed 7 marketing agencies including Credo and BBK Foundation.

AI AutoQA at the Coach Summit: Presenting to 80+ Enterprise Leaders

Presented AI AutoQA at enterprise summits in Barcelona and Seville.

Scaling QA with AI: From <1% to 100% Coverage

How a European telecom operator transformed QA operations with AI calibration at scale, reaching 82% accuracy while analyzing every customer interaction.

Micro-SaaS Sold to Factorial: Building ERP Integrations

Built a micro-SaaS that automated Italian ERP integrations for Factorial's HR platform. Sold the product to Factorial.

FlexRent: From Zero to 600 Members and Angel Investment

Founded a room exchange platform for digital nomads. Pitched at South Summit, Warsaw, DES. Featured on TV.

137% Engineering Velocity: Transforming Delivery at an HR Fintech

Redesigned delivery processes at Hastee, increasing engineering output 137%. Led 13 engineers across 3 teams.

DES Malaga: Pitching to Investors at the Digital Enterprise Show

Pitched FlexRent at one of Europe's largest digital transformation events.

Radio Interview: Technology and Entrepreneurship

Radio interview discussing technology, entrepreneurship, and innovation.

On Stage at South Summit: Europe's Leading Startup Conference

Main stage pitch at South Summit, one of Europe's top startup conferences.

Featured on TV: Discussing Tech Innovation on Hola Andalucia

Television appearance on 101 TV Malaga's tech segment.

Video Interview: Product Vision and Startup Execution

In-depth interview on product vision, market strategy, and startup execution.

Pitching at Warsaw Startup Stage

International startup pitch competition in Warsaw, Poland.

Why We Pivoted a $2M Trading Engine to Microloans

The story of how Xcapit evolved from 400%+ returns to UNICEF-backed financial inclusion. What I learned about building products that matter.

400% Returns: How Bayesian Optimization Tuned Our Trading Strategies

Gaussian Process-based hyperparameter optimization for algorithmic trading. Featured in Cronista and CriptoNoticias.

Building AI Before ChatGPT: Custom ML at a UNICEF-Backed Fintech

Neural networks, genetic algorithms, and Bayesian optimization built from scratch with a team of 5 PhDs. Before the LLM revolution.

Reducing Data Requests to Zero: Building a Self-Serve Analytics Bot

Discord bot that democratized data access across the company. Eliminated 100% of ad-hoc data requests.

Building an FX Arbitrage Engine: How I Automated a Trading Desk

Multi-threaded Python system that exploited FX pricing inefficiencies at Argentina's largest broker. Led to promotion to Head of BI.

21 Scrapers, 20% Market Share: Building a Competitive Intelligence Engine

How I built 21 real-time scrapers (including mobile app emulators) to monitor every competitor in the Argentine FX market.

From 5 Hours to Zero: Automating a Trading Desk's Post-Market Operations

Python-based automation that eliminated 5+ hours of daily manual work for a 5-person team. The evolution from VBA to full ETL pipelines.

Dealing with Heteroscedasticity

Technical article on statistical modeling and handling non-constant variance in financial data.