MIT MIT Capstone Project Worked with Comcast on real-time churn prediction from call transcripts using NLP & LLMs Sherlock Picasso Marketing made easy. We built a product to help SMBs attract customers on social media. AI-Powered Email Assistant for CMA CGM We built CMA CGM's AI-powered email assistant that generates recommended replies for over 100k weekly customer emails. What Videos Do You Love To Take? Video classification project using deep learning, combining Recurrent and Convolutional Neural Networks. Uber Data Exploration Analyzing and visualizing the data from Uber drives in New York City. Earnings Calls Processor Leveraging Machine Learning and NLP techniques to create a novel earnings calls dataset for stock return prediction Predicting Songs Popularity We built an interface to predict the popularity of Spotify songs based on their audio features. Prescribing Diabetes Classes We developed a tool to prescribe diabetes self-management classes to patients using optimal trees methods from survey data. Optimizing Bus Stops Selection We built optimization models to optimize the selection of bus stops for the MBTA (Massachusetts Bay Transportation Authority) by balancing profit maximization and efficiency of the bus network. Analytics for BigTechs in 2023-2024 A closer look at how the biggest tech players are using analytics to advance their key priorities. Personal projects AI Earth Hackathon By combining innovative LLMs and NLP, this tool enhances evaluations, guiding decisions on the effectiveness and viability of diverse solutions in the context of circular economy-related problems. Transformers Insights Exploring how Transformers actually transform the data under the hood. Check the associated Medium article for more details! Stable Diffusion from Scratch! An implementation of the Stable Diffusion algorithm and UNet from scratch to generate images using Python and TensorFlow. VAE from Scratch! A simple implementation of a Variational Autoencoder from scratch using Python and TensorFlow - check also the blog post! NER from Medical Documents Perform Named Entity Recognition (NER) on medical documents and medical literature, and annotates the PDFs with the recognized entities. Chess Engine A chess engine, written in Python, including a GUI and an agent that the user can play against. My First AI Project This is my very first AI project, in which the goal was to develop an interface to recognize handwritten text.
