A selection of projects that showcase my journey as a Full-Stack Developer. From developing modern websites to creating digital applications, each project stems from the goal of transforming an idea into a concrete, functional, and meticulously crafted solution.
Through these jobs, I explore front-end and back-end development, experimenting with modern technologies and an approach focused on product quality, user experience, and the continuous growth of my skills.
Full-stack web application for creating recipes, cooking step by step, and managing your shopping list
MamaChef was born from a university project of Software Design and Development, initially developed with two colleagues who later gave me the opportunity to carry on the project independently. The project was then migrated to Django and completed in a first phase, becoming the final submission for the CS50 Web course.
Design Prior to development, a requirements-gathering and design phase was conducted, based on needs identified by potential users.
Features resulting from this phase include recipe search by ingredient, a guided cooking mode, and the ability to add a recipe's ingredients directly to one's shopping list.
Ingredient Validation with NLP Adding new ingredients is one of the areas where I have been able to apply my data science knowledge most directly.
Each ingredient undergoes various levels of validation to ensure consistency with the ingredients already in the database. The system can identify very similar names, suggest matches, and prevent duplicates.
When an ingredient is not already present, the system verifies that it is indeed a food item and normalizes it linguistically through lemmatization to ensure data consistency within the database.
Full-stack Implementation The application's architecture integrates a complex relational data model, Django-based backend logic, and JavaScript-driven frontend interactions. I also developed features requiring coordination across the application's various layers, such as a guided cooking mode with timers, ingredient autocomplete, asynchronous shopping list updates, and PDF export.
Python library for word embeddings visualization and analysis, useful in Natural Language Processing
WordViz is designed to make a common activity in embedding analysis more accessible: exploring their structure through similarity, clustering, and dimensionality reduction.
The library consolidates several steps that would normally require separate implementations into a simple interface, transforming them into interactive and customizable visualizations accessible through a few lines of code.
Simplicity WordViz handles internally operations such as dimensionality reduction, clustering, and distance calculation, allowing the user to focus on exploring their own embeddings.
The visualizations are configurable through dedicated metrics and parameters, maintaining a balance between ease of use and control over the analysis.
Universal Loading The EmbeddingLoader object simplifies the loading of embeddings, handling both downloads and direct file loading while supporting various formats.
Once loaded, the embeddings can be passed directly to the Visualizer, without requiring additional preparation steps.
Dedicated Visualizations WordViz offers visualizations specifically designed for embedding analysis, including exploration of most similar words, similarity and distance heatmaps, dendrograms, and two- and three-dimensional representations.
The library also includes tools for customizing the appearance of visualizations through predefined palettes and themes, as well as features for managing more crowded plots, such as automatic label placement.
An interface will be developed to allow users to explore WordViz's features without directly using Python.
NOTONLYDESK, or NOD, is an innovative startup that offers a flexible platform to access a network of coworking spaces throughout Italy and Europe.
I collaborate with them, supporting the CTO during a phase of growth and product development, handling activities that range from full-stack development to technical SEO, up to data analysis and machine learning.
Full-stack Development I contributed to the development and evolution of the platform, working on both the frontend and backend. I created several static pages, developed the External Booking System in Django, and built much of the new management system for coworking space administrators.
I also took care of technical decisions, maintenance, and necessary interventions during the transition of the management system to production and the sales phase.
Technical SEO Within a dedicated subteam, I handle the technical SEO of the website. The work includes analyzing and resolving indexing issues, implementing structured data, improving accessibility, and evolving the site towards a more AI-friendly navigation.
Machine Learning My background in statistics has allowed me to contribute to the analysis of booking data. I have started an activity of predictive analysis that begins with data extraction, cleaning, and exploratory analysis, and will continue with the definition and validation of the model.
Interactive web application for statistical data analysis, machine learning modeling, and model explainability
StatPilot is designed to offer a completely no-code environment for quickly performing small data science tasks, without sacrificing the opportunity to delve deeper into the data and models used. The app guides the user from uploading a dataset to its analysis, up to the training and interpretation of machine learning models, through a simple and lightweight workflow.
Data Analysis The user can upload a dataset and obtain an initial exploratory analysis without preliminary configurations. StatPilot automatically identifies the type of variables, numeric, qualitative and textual, and adapts the available analyses accordingly, returning descriptive statistics, distributions and correlations both in numeric format and through visualizations.
Machine Learning The Machine Learning section allows users to choose from three classification and three regression models, select the target variable, and train the model directly from the interface.
In addition to evaluation metrics, the model can be used on new observations to obtain predictions.
Explainable AI StatPilot uses SHAP to analyze model behavior at different levels.
Global analysis allows users to observe the importance of variables in the overall model behavior, while local analysis enables the explanation of a single prediction by specifying the index of the observation to be analyzed.
Dataset Validation Making the application usable with different datasets required handling numerous cases related to the structure and quality of the input. I developed an input validation system that checks the data provided by the user and intercepts conditions that could compromise the subsequent phases of analysis or modeling.
Space Admin is a custom management platform developed for the administrators of NOD coworking spaces.
The panel allows for comprehensive space management, from bookings and workstations, to CRM in the free version, sales pipeline, and contract creation for the Pro plan.
The panel is not publicly accessible, but a demo can be requested by contacting NOD directly.
Frontend I built much of the management panel, translating the defined UX/UI foundations for the product into frontend implementations and addressing technical solutions when doubts or issues arose during testing phases.
Backend Beyond CRUD logic, I contributed to transforming business flows into application features through Django's ORM.
In particular, I developed the logic for the Pipeline and Contracts sections, enabling the tracking of a lead from the initial contact to the potential activation of a plan and managing the associated information, including dates, amounts, users, customized services, and reminders.
Maintenance I currently follow the maintenance of the management system, fixing bugs and introducing changes that have emerged from initial tests with users.
The work is still in progress, with the goal of preparing the product for sale.
Website for a photographer with a strong passion for supercars and motors in general
Redesign of the old website for a gynecologist specialized in hysteroscopy in Palermo, Italy
Interactive web application that, given a Spotify playlist or a list of 5 songs, can tell you what kind of listener you are
Showcase website for a university professor, presenting his work and the courses he teaches
Exploratory data analysis of Amazon food reviews, with sentiment analysis using Python and NLP techniques
Website for a no-profit organisation protecting sharks and all sea life in general
Dynamic platform for tracking cryptocurrency trends and market insights in real time
Website for a charming 4 stars hotel nestled in the heart of the Tuscan Maremma, Italy
Whether it's a new web application, a modern website, or a technical collaboration, I'm always excited about new challenges.
© 2026 Elena Zen All Rights Reserved
Privacy Policy - Cookie Policy