Greater Barcelona Metropolitan Area
I build AI systems and lead the teams that deliver them. My work sits at the intersection of machine learning research and enterprise product delivery — designing generative AI pipelines, deploying LLM and NLP integrations at scale, and translating complex AI trade-offs into decisions that business stakeholders can act on. At Colt Technology Services, I deliver an AI-powered alarm classification platform ingesting real-time telemetry from multiple network monitoring systems, applying probabilistic and deterministic AI models to automate triage and routing at scale. I own the full product lifecycle — from architecture decisions and model evaluation to cross-functional delivery and senior stakeholder alignment. My research benchmarks GAN architectures (TimeGAN, QuantGAN, FinGAN) against LLM-based generation strategies for synthetic financial time series on NVIDIA and BRICS equity data. Applications include data augmentation, privacy-preserving analytics, and model back-testing — with findings being prepared for journal submission. Academic background: MSc in Data Science (Barcelona School of Economics · UPF / UAB), Executive MBA in Corporate Finance (FGV), MSc in Materials Engineering and Nanotechnology (Mackenzie / Politecnico di Milano). Stack: Python · PyTorch · TensorFlow · LangChain · Transformers · RAG · Agentic AI · NLP · GANs · Azure · GCP · SQL · Power BI · Agile / Scrum German national based in Barcelona. Actively targeting AI, data science, project manager and product roles in Germany. → [email protected]
- Architecting an AI alarm classification system ingesting real-time telemetry from SMARTS, NETCOOL, and Skylight; automated routing reduces manual network operations handling at scale. - Designing and deploying NLP and LLM pipelines for automated customer communication flows, including prompt engineering, output validation, and API integration with external partners. - Coordinating cross-functional agile teams across data engineering, AI/ML, network ops, and external vendors; serving as Scrum Master and Product Owner across concurrent workstreams. - Delivering on Azure cloud infrastructure; enabling RPA integrations and managing budget alignment across parallel AI initiatives.
- Built an end-to-end Retrieval-Augmented Generation (RAG) system over SEC EDGAR filings, enabling natural language querying of financial disclosures, 10-K reports, and earnings data. - Developed edgar_fetcher.py — a custom data pipeline for programmatic ingestion, parsing, and chunking of EDGAR documents for vector store indexing. - Deployed as a live application on Hugging Face Spaces via CI/CD pipeline on GitHub, enabling automated builds and continuous deployment on every commit. - Stack: Python · LangChain · RAG · Vector DB · HuggingFace · GitHub Actions · CI/CD
- Implemented and benchmarked three GAN architectures (TimeGAN, QuantGAN, FinGAN) and three LLM generation strategies (zero-shot, few-shot, fine-tuned) for synthetic financial time series on NVIDIA and BRICS equity indices. - Designed an integrated evaluation pipeline assessing fidelity, diversity, and statistical properties of synthetic vs. real return distributions; results being prepared for journal submission. - Direct applications: data augmentation, model back-testing, privacy-preserving analytics, and regulatory compliance in financial services and healthcare AI.
• Responsible for issuing purchase orders • Manage the project's budget • Manage all departments involved in the process