Built and deployed a full-stack data application processing 150+ years of infrastructure records, forming the basis of a project projected to generate millions in revenue.
Developed an AWS-native agentic system that automates the full data migration lifecycle: inferring and enriching dataset metadata, aligning schemas to the Swisscom Enterprise model, and generating ready-to-use PySpark/PyFlink jobs — significantly reducing manual engineering effort.
Designed, implemented, and trained a Spatio-Temporal Graph Neural Network to improve traffic prediction and network energy efficiency. Built the full data pipeline using PySpark and Airflow to structure network telemetry as graphs and time series, jointly capturing spatial topology and temporal dynamics.
Onboard and mentor new graduates in agile practices, DevOps, and productization, accelerating their integration into the team.
Nominated to the Swisscom Talent Program, reserved for 2-3% of the company. Selected to receive dedicated training in leadership and soft skills development.
Master thesis at the Swisscom Digital Lab. Achieved a 2% improvement in RAG accuracy by implementing token-level classification between domain-specific (Swisscom) and irrelevant tokens.
TA across two courses: MATH-232 Probabilities & Statistics (250+ students) and Statistics for Data Science master course (150+ students). Question answering, exam invigilator and corrector.
Acted as Product Owner for 6-student agile teams, running weekly sprint ceremonies, unblocking technical issues, and evaluating deliverables across software engineering and design.
Covered deep learning architectures (CNNs, RNNs, Transformers), graph neural networks, and large-scale NLP. Complemented by rigorous statistical inference, Markov chain modelling, and applied projects in data analysis, visualisation, and computer vision.
ID: 0067585
Verify diploma →Core curriculum spanning Fourier analysis, signal processing, digital communications, and stochastic processes. Mathematical foundations in calculus, probability theory, and statistics, complemented by applied courses in machine learning and software engineering.
ID: 0058803
Verify diploma →One-year preparatory programme at EPFL for students from vocational backgrounds, bridging applied sciences and the academic requirements of the bachelor's degree.
Practical curriculum covering object-oriented programming, relational database design (SQL), full-stack web development (HTML, CSS, JS, PHP), and network protocols and infrastructure.
Vocational baccalaureate (orientation technique) complementing the CFC apprenticeship, providing the qualification for entry into Swiss universities of applied sciences.
Personal experiments, open-source work, and academic contributions
Trained reinforcement learning agents to play Jass, the popular Swiss card game. Implemented and compared multiple RL strategies to develop competitive game-playing policies from self-play.
Explored deep learning applied to audio data, including feature extraction, classification, and generation tasks. Worked with spectrograms and waveform-based models to understand the specifics of audio as a modality.
Developed a deepfake video classifier leveraging temporal information from H.264 motion vectors to outperform frame-level classifiers. The approach enables potential real-time detection with minimal computational overhead. Led to a publication at EI'2024.
Interactive D3.js data visualisation of the Harry Potter universe. Involved data processing and merging of multiple datasets, and text analysis using scikit-learn to extract patterns across the series.
Developed an event management platform for the EPFL community, enabling student associations to organise, publish, and manage events on campus.
Efficient Temporally-Aware DeepFake Detection using H.264 Motion Vectors
arXiv · Computer Vision and Pattern Recognition · 2023
Proposes leveraging motion vectors and information masks from H.264 video codecs to detect temporal inconsistencies in deepfake videos with minimal computational overhead, enabling potential real-time detection.
View on arXiv →Amazon Web Services · Feb 2024 - Mar 2027
Foundational understanding of AWS Cloud services, architecture, security, and billing.
ID: 96b1266a27c043578ee736e260e400ee
Amazon Web Services · Mar 2024 - Mar 2027
Designing distributed, scalable, and fault-tolerant systems on AWS. Architecture best practices and cost optimisation.
ID: d19c8d7aa392433194b0e55b53fb3d0a
Amazon Web Services · Nov 2025 - Nov 2028
Foundational knowledge of AI/ML concepts, AWS AI services, responsible AI practices, and generative AI on AWS.
ID: 7a19690e7a3c44c19861fd3d7ca669ca
The Linux Foundation · Nov 2025 - Nov 2027
Designing, building, configuring, and exposing cloud-native applications for Kubernetes.
ID: LF-194oglwx4w
What I do outside of work
FC Les Bois · Tartare de Miettes · Balelec · 2016 - Present
Event organizer and IT Manager for a football club, a music festival, and Balelec — handling events with 1,000+ attendees, volunteer teams, web development, and account management.
Personal interest
Enjoy a wide range of genres — open-world exploration, narrative-driven adventures, and card-based strategy games. Also a fan of board games: card, strategy, and cooperative.
Self-directed
Keep up to date with the latest developments in ML, AI, and data science by reading research papers and documentation, watching courses, and following the field for both personal culture and professional growth.
Personal interest
Love to travel across the world to discover new cultures and places. Enjoy hiking to reach stunning landscapes and capturing them through photography.
