SOFTWARE ENGINEER · SPAIN
I'm Txema Bermúdez, a Senior Software Engineer with 10+ years building backend systems, APIs and distributed architectures, primarily with Python and FastAPI. I'm currently expanding into AI engineering and generative AI.
01 / PROFILE
I enjoy solving complex problems, making architectural decisions and turning ambiguous requirements into simple, maintainable systems.
I have worked across startups and product companies, often owning systems from initial design through production and long-term evolution. My experience spans scalable backend services, asynchronous and distributed systems, data pipelines, ETL, cloud platforms and data-intensive applications. In healthcare and AI products, reliability, data quality, scalability and regulatory requirements have been central.
CURRENT FOCUS · AI ENGINEERING
I'm expanding my focus towards LLM applications, RAG, tool calling, agents and AI infrastructure. I use AI tools regularly in my development workflow—as a way to move faster while keeping engineering judgement, code quality, testing and reliability at the centre.
02 / EXPERTISE
03 / CAREER
04 / EDUCATION, PUBLICATION & PROJECTS
University of Huelva · 2007–2009
Final grade: 8.25 / 10
University of Huelva · 2000–2007
Final grade: 7.76 / 10
Proceedings of the Workshop on Challenges in Natural Language Processing for Clinical Data (November 2008). The work studies negation handling in clinical narratives and the use of controlled vocabularies to reconcile different expressions of the same medical concept.
An open-source personal finance platform focused on expense tracking and financial management.
AsGeRe is a Java desktop application I built as my final degree project in biomedical NLP. It supports the creation of manually written extractive summaries, or gold-standard summaries, used as reference material to evaluate automatic summarisation systems.
Users can load PDF, RTF, TXT and XML files, paste text or retrieve articles from BioMed Central. The app extracts and segments the text, detects its language and applies language-appropriate processing: MetaMap and UMLS concept extraction for English, and stop-word removal and stemming for Spanish. Users select sentences to build a summary, compare word and sentence counts and concept-frequency distributions between the source and summary, then save documents and summaries in MySQL for reuse.
A self-hosted homelab running personal infrastructure, development tools, vehicle telemetry, media and home automation services.
