EN DE ES

Juan-Francisco (Paco) Reyes

AI Engineer | PhD Candidate

AI Engineer and NLP researcher with a background in Human-Computer Interaction and full-stack development. I specialize in building machine learning and generative AI solutions powered by Large Language Models (LLMs) and have hands-on experience in MLOps for deploying these models. I design and develop AI products that bridge human-machine communication with a focus on automation, and explainability, combining computational linguistics with solid software engineering and user-centered design.

➔ Download CV (PDF, 134 KB, 2 pages)

Professional Experience

Academic Researcher & Lecturer

Brandenburgische Technische Universität (BTU), Cottbus, Germany

April 2020 — March 2024

  • Led research on explainable AI (XAI) in NLP, developing three models and datasets published on Hugging Face, enhancing transparency in LLM-based political text analysis, resulting in the publication of three peer-reviewed papers on computational linguistics.

  • Designed and lectured the NLP for Information Extraction course, integrating tools like spaCy and Hugging Face Transformers, resulting in an increase in student enrollment over four years.

  • Supervised two Master's theses in NLP and Information Extraction and interdisciplinary student projects combining NLP and web development.

  • Co-authored the book Learn How to Build Web Apps with JS and Cloud Firebase, which focused on object-oriented programming (OOP) using serverless web technologies and the NoSQL Firestore. The book was used to lecture the course Web Applications, using JS, Node.js, Python, and Flask.

  • Lectured the course Web Documents, covering fundamentals and modern techniques in web documents, including HTML, XML, CSS, SEO, accessibility, performance, and others.

Founder, Director, Front-End Developer, UX Designer

Talámica Øptimization (UX agency), Lima, Peru

November 2014 — February 2020

  • Front-end development of marketing channels with a focus on user experience across diverse industries, including media, finance, banking, e-commerce, education, and startups.

  • Front-end development, primarily utilizing JavaScript, HTML, and CSS to enhance user interfaces through interaction design techniques, ensuring engaging and intuitive user experiences.

  • Orchestrated a team of web developers and UX designers, fostering a collaborative environment focused on web experiences and business goals.

  • Directed end-to-end UX design, from user research to implementation, using diverse methods (e.g., surveys, analytics, usability testing, prototyping) to develop strategic alignment and impact.

User Experience Director

Wunderman Thompson (digital agency), Lima, Peru

August 2013 — October 2014

  • Played a key role in client pitches by showcasing strong user insight, boosting client confidence, and driving sales up 30% in 2014 and 40% in 2015 for a total revenue gain of $500K+.

  • Recruited and led the UX team toward adopting an evidence-based approach to design, where decisions were driven by data and users' behavior.

  • Led a major initiative to revamp production processes, cutting time-to-market by 7% and boosting client satisfaction by 11%, embedding user-centered design in web and mobile development.

Graduate Assistant Researcher & Developer

University of Baltimore, Maryland, U.S.A.

September 2011 — July 2013

  • Developed Langsdale on the Go!, a serverless web application using HTML, CSS, and JavaScript, integrating multiple APIs to locate books in the library, featuring a book locator that displayed a map with the exact shelf where a resource was located.

  • Designed the interaction experience using core web principles and validated it through usability testing, interviews, and observation, improving patron satisfaction and increasing adoption rates.

  • Recognized by university authorities for pioneering innovation in digital library services and featured at the Internet Librarian 2012 Conference (Monterey, California, U.S.A.).

AI / NLP Projects

docuRAG

docuRAG

Retrieval-Augmented Generation Embeddings FastAPI LlamaIndex LlaMA 2 Docker Terraform Next.js LLM Generative AI Python JS

OntoLoop

OntoLoop

Embeddings FastAPI LlamaIndex Llama 2 Docker Terraform Next.js LLM Python JavaScript

webCrawler

webCrawler

Web Crawling Web Scraping NLP Flask BeautifulSoup Firebase

annotationNLP

annotationNLP

Django TensorFlow OpenCV

ObPolDis

ObPolDis

Information Extraction Knowledge Graph SVO triples BeautifulSoup Web Scraping JavaScript Python spaCy Firebase Eleventy

ieApp

ieApp

Information Extraction Web Scraping Neo4j BeautifulSoup

Text Classification

Text Classification

Computational Linguistics NLP BERT Dataset Explainable AI LLM SHAP Hugging Face Python spaCy Optuna

Stance Classification

Stance Classification

Computational Linguistics NLP SetFit LLM Explainable AI Dataset SHAP Hugging Face Python Firebase Embeddings

Topic Continuity Analysis

Topic Continuity Analysis

Computational Linguistics NLP BERT LLM Explainable AI Dataset Sentence Pair Modeling Transformers Interpret spaCy

Education

Skills

  • Development: PythonJavaScriptFlaskNode.jsSQLNoSQLOOPAPI DevelopmentWeb Design

  • AI and LLMs: NLPMLLLMsDeep LearningFine-tuningPyTorchSciKit-LearnSHAPLIMETransformer InterpretHugging FaceGenerative AIPrompt EngineeringOpenAI APILlaMARAGspaCyExplainable AILangChainLlamaIndexData ScienceStatistical AnalysisInformation ExtractionComputational LinguisticsMachine LearningKnowledge Graph

  • Cloud and MLOps: AWSGoogle Cloud PlatformFirebaseDockerPineconeFAISSNeo4j

  • User Research: UXUsability EngineeringInteraction DesignUser-Centered Design

  • Soft Skills: CommunicationPresentation SkillsAnalytical ThinkingTeam BuildingProject Management

  • AI IDE: Cursor AIJetBrains (Junie)PyCHarmWebStorm

Publications

  • AI Explainability in Classifying Political Speeches and Interviews

    Journal of Language Technology and Computational Linguistics (JLCL), under review | 2025

  • Explainable Subjective Stance Classification with SetFit in Political Discourse

    Journal of Language Technology and Computational Linguistics (JLCL), under review | 2025

  • Explainable Topic Continuity in Political Discourse: A Sentence Pair BERT Model Analysis

    International Journal of Computational Linguistics (IJCL) | 2025

Awards and Honors

  • First in a class of 23 students graduating, undergraduate studies at UNMSM, Valedictorian, (2006).

  • EducationUSA Opportunity Grants (Fulbright Peru), awarded financial aid for the application process to graduate studies in the U.S.A., (2007).

  • Graduate Assistant at Langsdale Library at University of Baltimore, full scholarship for Master’s studies, (2011).

  • Turner Research and Travel Award from the University of Baltimore (2012).

Certifications

  • AWS Certified Machine Learning – Specialty

    Amazon Web Services (AWS) | NOV 2025 (EXPECTED)

  • Certified Usability Analyst, HFI CUA™

    Human Factors International, credential #2010-2559 | 2010

  • Certified User Experience Analyst - HFI CXA™

    Human Factors International, credential #2015-327 | 2015

Languages

  • English: Full professional proficiency.

  • Spanish: Native.

  • German: B1 TELC | SEPT 2015 (EXPECTED).

About Me

With over 4+ years of experience in AI and machine learning, I’ve worked on real-world challenges through interdisciplinary projects in both academic and applied settings. My expertise spans machine learning engineering, NLP research, and AI product development, with a strong focus on model explainability 🧠 and NLP automation 🤖.

Personal Information