Experience

Staff engineering for LLM and agent systems.

I lead Data & AI at TaxDown and work directly on architecture, implementation, evaluation, reliability and production operations.

2M+ customer-facing LLM interactions in production

~3M automated support messages, 2026 YTD

~96% accuracy in human-reviewed production audits

~80% CSAT on automated customer support replies

taxdown.md

TaxDown

I am Head of Data & AI and a member of TaxDown’s Executive Committee, responsible for technical direction across AI and data. I also contribute directly to architecture and implementation.

I built Adri, a multi-agent system used in production for tax operations. It processes documents, extracts structured data, updates tax returns, diagnoses errors and routes selected cases to human review.

production-ai.md

Production AI

I design and operate agent systems with RAG, structured outputs and tool calling, using frameworks and SDKs such as LangGraph and the Vercel AI SDK. Evaluation and reliability work covers tracing and observability, evaluation datasets, human audits, guardrails and fallback across models from OpenAI, Anthropic and Google.

These systems run in production across tax preparation, document processing, payments and customer support.

engineering-background.md

Engineering background

Before specializing in AI, I worked at Expedia, Infinia, SpringCM, Channel IQ and HERE/Nokia on data platforms, streaming systems, cloud migrations and distributed backend services.

Education: Telecommunications Engineering at Universidad Politécnica de Madrid and M.S. in Information Technology & Management at Illinois Institute of Technology.