I'm Pol Resina, a Data Scientist with a passion for turning data into actionable insights. Currently, I'm studying Data Science & Engineering Bachelor's Degree at the Universitat Politècnica de Catalunya (UPC).
Here are a few things I've been working on:
I've been working on a project with the UPC Formula Student Driverless team. We are developing a racing car to compete in the Formula Student competition. The car must be able to drive autonomously and complete a series of challenges. I'm in charge of the perception department: Using a LiDAR we have to make sure the car can detect cones on the track and immediately get the track limits. Click here to see the perception pipeline in action.
Overview: Ros package that detects cones in the track using a LiDAR. It's based on DBSCAN clustering algorithm to group the points that belong to the same cone. What's more, few optimizations have been added to improve driverless car's performance. The system is able to detect the cones and publish the position of each cone in the track. Compensation for the raw pointcloud data has been added to improve the detection of the cones. Please ensure you select the correct branch if you wish to view the code, as the master branch is empty.
Tech used: C++, ROS, PCL
Keywords: Machine Learning, Optimizatinos, PCL, clustering, DBSCAN,
Overview: Machine learning project to predict the severity of car accidents in the US. We have used a dataset with over 3 million records to train a model that can predict the severity of an accident based on the weather, road conditions, and other factors. Also, there is an intoductory part of data analysis and visualization. A report written in catalan is attached.
Tech used: Python, Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn.
Keywords: Data Analysis, Data Visualization, Machine Learning, Classification.
Overview: Developed a solution to identify duplicate or similar images within a dataset and built a clothing recommendation system that suggests outfits based on the user's input. Additionally, I developed an assistant system to help Inditex manage out-of-stock issues. For these tasks, I leveraged the pre-trained CLIP model to understand image context and used a U-NET model for semantic segmentation to detect and segment clothing items in images.
Tech used: Python, Pytorch, Huggingface, CLIP, U-NET, Semantic Segmentation
Keywords: Image Embedding, Contextual Understanding, Clothing Recommendation, Semantic Segmentation, FrontEnd
Video Demo: Street Style Decoder
Overview:A program designed to derive the optimal football team composition from a collection of current players. It's mainly crafted with C++.
Tech used: C++, Python
Keywords: Optimization, Algorithms, Data Structures
Overview: A desktop application that allows individuals to explore available movie showtimes and offers directions from their current whereabouts to the cinema, using the actual bus network in Barcelona.
Tech used: Python, Tkinter
Keywords: Optimization, Algorithms, Data Structures, GUI
Overview: Using a combination of Genetic Algorithms and Simulated Annealing, we have devised a resolution for a modified version of the Traveling Salesman Problem.
Tech used: Python, Matplotlib
Keywords: Optimization, Algorithms, Data Structures, GUI
Overview: My first project in Python. It's a password manager which is a software tool that securely stores all your passwords and other sensitive information in one place. It uses a master password to encrypt the data. The use
Tech used: Python, SQLite, Tkinter
Keywords: Data Security, GUI
- Programming Languages: Python, R, SQL, C++, C, Java, Matlab, Shell
- Data Analysis Libraries: Pandas, NumPy, Polars
- Data Visualization Tools: Matplotlib, Seaborn
- Database Management: MySQL, SQLite, DBeaver
- Machine Learning: Pytorch, TensorFlow, Scikit-learn
Feel free to reach out to me on LinkedIn or send me an email.
I'm always open to discussing data science projects, collaborations, or job opportunities. Let's connect!