I’m Raúl Ferrer — a software architect focused on AI systems reliability.

Most AI systems today are not engineered — they are assembled. They work in demos, but fail in production due to poor evaluation, lack of observability, and fragile architectures.

This site explores those failures.

I analyze how AI systems behave under real-world conditions:

• Noisy and evolving data.
• Latency and cost constraints.
• Non-deterministic outputs.
• Degradation over time.

The goal is simple:

Understand why AI systems fail — and how to design them to be reliable.