My Work

A glimpse into my recent work, where you'll find a mix of my own projects and client collaborations. Blending creativity and functionality, each project was created with care, expertise and commitment to excellence. Take a look around and get inspired.

Portfolio & Tutorials

This very website, along with its blog engine, has been meticulously crafted and developed with a host of dynamic functionalities. Functionalities that empower users to seamlessly add, delete, and modify posts directly from the front end, ensuring a user-friendly experience. Furthermore, the platform facilitates collaboration by enabling other users to publish their own posts. Also, readers may engage with the content through comments, although this feature is presently inactive.

Hospitality Website

WORK IN PROGRESS: A website for a charming boutique bed & breakfast. The website incorporates seamless integration of a booking system, alongside user-friendly functionalities that allow the client to autonomously update and tweak content without necessitating my direct involvement. The website will be proficiently multilingual, catering to a diverse audience with translations available in English, Italian, French, and Spanish.

Digital Transformation

I collaborated with a major player in the Spanish energy market to digitalise a manual process that heavily relied on Excel spreadsheets and Word documents. Previously, this monthly task demanded a full workweek and involved multiple team members. Through the web application I developed, the entire process now takes just around 40 seconds, a substantial improvement in efficiency. Due to confidentiality requirements, the web app is accessible behind a login page. However, I can arrange a demonstration with simulated data if needed.

Scraping

Tecla is a powerful web scraper designed to efficiently crawl URLs in bulk. It accurately extracts text from web pages and documents, offering the flexibility to upload your own documents in large batches. The resulting output is conveniently downloadable in JSON format, readily compatible with various data analytics tools, including Excel.

Natural Language Processing

Lexi is an AI-powered web application engineered for information extraction from unstructured text, and document clustering. It doesn't rely on a GPT wrapper; instead, it was developed from the ground up, centered around fine-tuning a DistilBert model. I personally enriched the training dataset by manually annotating approximately 5000 sentences, thereby expanding the scope of entity groups, including specialized categories like medical and legal terms. Additionally, the web app offers a feature to annotate custom entities. Lexi can analyse various document formats such as PDFs, Word, TXT, and XLSX files in bulk. Moreover, it effortlessly handles documents scraped from the web, allowing users to input URLs for extraction.

Open Source

Nerval is an open-source package, accessible on both PYPI and Anaconda platforms. Its primary function is to assess Named Entity Recognition (NER) models, generating entity-level confusion matrices and classification reports. I developed it in 2022 out of the necessity for a tool tailored specifically for NER evaluation. Recognizing the gap in the solutions available at the time, I took it upon myself to create a dedicated package. I believed that other NLP practitioners would find it useful, therefore I opted to release it as open source.