Nour Ghribi

Visiolinguistic Image Search

Master Semester Project: Researching Multi-Modal framework for article image retrieval and matching:
- Reviewed and extensively researched State-Of-the-art Machine Learning models to replicate and combine them in a creative setting.
- Concluded the project by implementing a proof-of-concept on fashion data that leverages interactive image retrieval, keyword extraction, and image clustering.

Movie Recommender

This project was during the course Systems of Data Science which covers the important notions necessary for the deployment of todays state of the art data science solutions. What made me proficient in Scala and Spark without a doubt. I learned how to make use of modern clusters and cloud offerings to scale up to very large workloads and analyze the trade-offs between various approaches to large-scale data management and analytics. This helps to choose the most appropriate existing systems architecture and technology for a task. - Developed a movie recommendation system leveraging user ratings and similarities for clustering from publicly available data and scaled it up to work on large databases using Spark and Scala. - Concluded the project by evaluating different system deployment strategies' price/performance.

Predicting the "Present" with Google Trends

Who has ever watched movies or documentaries about the 2008 crisis like "The Big Short" or "Inside Job" and wondered "If only I could predict the stock market crisis and make so much money by betting against the market?" We did.
- A replication of the paper Predicting the Present with Google Trends followed by a creative extension.
- Leveraged Auto-Regressive models improved by time series data of Google search engine trends to predict index volatility in financial markets.
My team and I used the techniques proposed in the paper to devise a model for market volatility predictions and tested it on the Dow Jones Index to assess it.

NLP Sentiment Analysis - Tweets Analysis

Where I found my passion for NLP before the rise of LLM. A practical use of Natural Language Processing is sentiment analysis. This project aims to devise a sentiment classifier for tweets in English. During the project, I implemented different machine learning classifiers (Linear models and Deep Neural Networks) to detect positive and negative sentiments in tweets. GloVe embeddings and famous state-of-the-art NLP models such as ELMo and BERT helped make this project an astounding success. Learned to benchmark different machine learning models to detect positive/negative sentiments in tweets, by using basic linear models and sophisticated deep learning ones.

Homemade PyTorch Framework

Deep Learning: Under the hood of Neural Networks with PyTorch. Discovered how deep learning framework PyTorch works by re-implementing some modules from scratch and learned the logic used and the power of underlying mechanisms that helped me develop an ease of working in PyTorch and deep learning.
Implemented the modules of pytorch for sequential neural networks from scratch (Linear layers, Activation Functions, Initializers, forward propagation, backpropagation and optimizers).
Benchmarking several neural net architectures combining Siamese Networks with weight sharing and auxiliary losses to predict a comparison between two hand-written digits from the MNIST dataset using PyTorch.

Robust Journey Planner

Improved SBB-CFF journey planner by considering historical success rates using statistical modeling. Dealt with Big Data: all the SBB-CFF lines timeltables, using Spark and Hadoop.
Modelled the data in a graph using NetworkX.
Modelled the delay of each line probabilistically.
Created a routing model that prioritizes success of connection instead of shorter travels.

Blockbuster Movies Recipe

What if production companies and publishers had a magic recipe for Blockbuster Movies? Choosing the right cast members and executive team such that they are guarenteed a prominent success? Through thourough analysis of data (Exploratory Data Analysis) and fitting an appropriate ML model, we tried to predict movies success (ratings and revenues) and find the magic recipe for the greatest grossing film.

Hello, Bonjour, Grüezi, مرحبا 👋 I'm Nour Ghribi, Senior Data Scientist & AI Engineer at Artefact (ex-Effixis).

My Story

Technical Skills

Generative AI

Programming Languages

ML/AI Tools

Data Tools

Development Tools

Cloud Platforms

Professional Experience

Senior Data Scientist & AI Engineer

Artefact (ex-Effixis)

Data Scientist Intern

Compagnie Financière Tradition