AutoML-Engineering
About
Motivated by an eclectic curiosity for Computer Science and a knack for innovation. I pursued a software engineering B.Sc. and dual M.Sc. advanced computer science joint machine learning degrees. This academic path propelled me into the exciting world of research, where I found myself, primarily due to the ability to push the boundaries of what is possible by uncovering previously uncharted areas and influencing the direction towards world is headed. My multifaceted experience—ranging from agile startups to R&D departments in larger corporations—has honed my expertise, positioning me at the forefront of machine learning innovations for tomorrow’s landscape.
Projects
Auto-Scikit-Longitudinal (``Auto-Sklong’’) is an automated machine learning (AutoML) library designed to tackle the Combined Algorithm Selection and Hyperparameter Optimization (CASH Optimization) of the Longitudinal ML classification task using various search methods. Including, Bayesian Optimisation via SMAC3, Asynchronous Successive Halving, Evolutionary Algorithms, and Random Search via GAMA.
Scikit-longitudinal (``Sklong’’) is a machine learning (ML) library developed from scratch designed to tackle the Longitudinal ML classification task. It offers tools and models for processing, analysing, and predicting longitudinal data, with a user-friendly interface that integrates with the Scikit-Learn – “Fit”, “Predict” “Transform” – ecosystem.
Speaking
In this talk, we discuss our experience with GitHub Copilot, an intelligent peer programming tool (AI pair programmer). We will look at how to use it, the benefits it provides, and the limitations we have discovered. In doing so, we will try to provide some hints as to whether this plugin delivers on its promise of allowing developers to write code faster and with less effort. We will begin with a live-coding micro-session to track suggestions in real time and compare them from user to user. We'll then look at the strengths and questions that arise when using the tool. Finally, we'll conclude with a short discussion of how GitHub Copilot is changing the way developers document their code.
Writing
Conference : International Conference on Bioinformatics and Biomedicine (BIBM) 2024.
Contribution: (I) A novel sequential search space addressing the Longitudinal Machine Learning Classification task – ``Auto-Sklong’’ (II) A brand-new toolkit set for Longitudinal Machine Learning analysis out of the box – ``Scikit-Longitudinal’’ (III) Statistical evidence that extending the temporal information inherent in longitudinal datasets in longitudinal machine learning leads to higher predictive accuracy.
Authors: David Tighe, Jeremy McMahon, Clare Schilling, Michael Ho, Simon Provost, Alex Freitas
Contribution: Main Machine Learning Researcher.
Authors: David Tighe, Jeremy McMahon, Clare Schilling, Michael Ho, Simon Provost, Alex Freitas
Contribution: Main Machine Learning Researcher.
Awards
Under a volunteer programme, AutoML's conference offered travel funding compensation for the entire conference experience.
I have been chosen to receive ESSAI & ACAI's complete summer school experience for going beyond the standard expectation of my Ph.D long-term’s perpsective.
Development of a risk adjustment model for positivity of surgical margins for Non-Melanoma Skin Cancer.
Development of a risk adjusment model for free flap outcomes.
One of the eighteen students selected from more than two thousand applicants to receive financial private scholarship to support their ambitious life project and academic pursuits.
Work Experience
Mission: Planning the Socials' aspect of the International AutoML Conference (AutoML.CC) edition in Paris under Prof. Carola Doerr's supervision, with teammates Dr. Elena Raponi and Dr. Anja Jankovic.
Teaching Assistantships: Introduction to (programming for) AI; E-Health; Object-Oriented-Programming.
Mission: Design and development of Intelligent Azure cloud-based solutions for a smarter recruiting process.
Mission: Research and Development of risk adjustment algorithms for Non-Melanoma Skin Cancers and Squamous cell carcinoma in oral and maxillofacial surgery with the use of AutoML and Imbalanced-based supervised ML algorithms.
Mission (Part-time): Design and data-architecture of Intelligent Azure cloud-based solutions for a smarter recruiting process.
Mission: Development of a cloud-based automated daily aggregation and pre-processing of French COVID-19 health statistics for HuffPost France journalists, enhancing their reporting efficiency.
Teaching Assistantships: Intensive C-based courses, Introduction Object-Oriented-Programming in C++, Epitech Innovative End-of-Year project supervision.
Mission: Development of a cutting-edge software by transferring a Windows-only Augmented Reality algorithm system to the Android/UNIX platform.
Volunteering
E-mma is currently nurturing the world's digital diversity. To define digital and show how computer science is open to all genders, and how mixity can boost a project. I have attended over 50 school interventions, student saloons, and roundtables. We worked closely with Google, the French State Secretariat for Digital Economy, Gender Equality, and Anti-Discrimination, the French Academy of Versaille, and Epitech.
Education
I architect cutting-edge Automated Machine Learning (AutoML) systems, uniquely engineered for unexplored (medical) data-types. Specialising in supervised learning within tabular-based data contexts, I leverage advanced optimisation paradigms—Bayesian Optimisation, Evolutionary Algorithms, and (Multi-)Armed Bandits—to develop algorithmic variants tailored for complex medical data types, such as e.g., Repeated Measurements.
Research Supervisor: Prof. Alex A. Freitas.
Dissertation: Classification of Sleep-Wake States Employing Lightweight Hardware and Machine Learning. Research Supervisor: Jean Noriot. Collaboration with: Prof. Yves DAUVILLIERS at the University Hospital Center Montpellier, France.
Advanced Computer Science (Computational Inteligence)
Dissertation: AutoML applied to medical data (Auto-Sklearn vs. Auto-WEKA). Research Supervisor: Prof. Alex A. Freitas. Collaboration with: Dr. David Tighe of the East Kent Hospital University NHS Foundation Trust.
Leadership Roles: Selected student organiser for supererogatory meet-ups featuring trending programming and research groups on campus.