My AI Research.
The core of my research focus lies in studying how to leverage non-Euclidean geometries inside AI architectures, in order to better represent data and its meaning. This ranges from hyperbolic or elliptic spaces, to matrix manifolds and Wasserstein geometries. Applications include word embeddings, language models, graph neural networks and optimization among others.
My Experience.
Gematria Technologies, cofounder/CEO
We served banks, asset managers and quant hedge funds with tens of millions of granular entity-level AI signals computed on news articles and customer reviews. Gematria also plugged on your in-house data feeds to create proprietary enriched datasets #NLP #AlternativeData #ESG #LargeLanguageModels #AutoGPT
AI Researcher
Working with Bernhard Schölkopf (the director of the institute) and Alexander Smola (global head of AI at Amazon Web Services) on Group Testing methods designed against COVID-19 #Bayesian #GraphicalModels #BloomFilters #Entropy
AI Researcher
Fully funded thanks to the DARPA & MLPDS, I worked with Prof. Tommi Jaakkola & Prof. Regina Barzilay on Graph Neural Networks for Drug Discovery. We built a new graph neural network architecture leveraging optimal transport geometry, with the aim to help discover new antibiotics #GraphNeuralNetworks #DrugDiscovery #OptimalTransport
PhD in Machine Learning
Thesis: On the Geometry of Data Representations. Advisor: Prof. Thomas Hofmann -- ex-Director of Engineering of Google Zürich. I wrote and corrected exams for master students in Deep Learning, Computational Intelligence Lab, Natural Language Understanding and I took the courses Cryptography Foundations and Cryptographic Protocols #Hyperbolic #MachineLearning #Representations #WordEmbeddings #LanguageModels #Optimization
Master in Mathematics
I was a member of the Queens' College and the Trinity Mathematical Society (TMS). Thesis: Statistical Applications of Persistent Homology, supervised by Prof. John Aston in Statistics and Prof. Jacob Rasmussen in Algebraic Topology #AdvancedProbability #DifferentialGeometry #StochasticCalculus #SchrammLoewnerEvolutions #MorseTheory #ModernStatisticalMethods