Description

Astrophysics researcher at Oxford with 10+ years experience with data analysis, statistical models, building ML pipelines and scalable computational tools for high-dimensional, multimodal time-series, imaging and hyperspectral data from large telescopes and simulations. Proven ability to extract meaningful patterns from complex, noisy data using Bayesian methods, deep learning and generative models. Highly experienced at applying theoretical and mathematical insight to a range of challenges with fast feedback loops and rich data visualisation. Excited to bring analytical rigour, creative problem solving, deep learning knowledge and coding excellence to AI/ML research roles in collaborative and fast-paced environments.

Langues

Anglais
Bilingue ou natif
Hindi
Bilingue ou natif
Italien
Notions

Préférences en matière de lieu de travail

Accepte de travailler sur site

London (jusqu’à 15 km)

UNIVERSITY OF OXFORD
POSTDOCTORAL RESEARCHER
CENTRES DE RECHERCHE
septembre 2022 - Aujourd'hui (3 ans et 9 mois)
Oxford, Royaume-Uni
• Developed a deep unsupervised learning model to classify spectroscopic/time-series data from James Webb Space Telescope, implementing a Variational Autoencoder (VAE) architecture achieving 98% reconstruction accuracy over varying noise properties. Performed clustering in the latent space to identify inputs with similar properties, discover edge cases and outliers, and generate realistic synthetic datasets from the embeddings. Accepted in NeurIPS Workshop 2025.
• Side project implementing autoencoder+GAN (TimeGAN) architecture to generate realistic synthetic US Treasury Bond yield curves for risk modelling. Project available on private GitHub repo upon request.
• Developed optimisation workflows leveraging Genetic Algorithms and parallel computing to extract signals from noisy spectroscopic/time-series data, achieving >20x speedup in optimisation tasks.
• Released production-grade Python APIs for data sanitisation, masking, and robust model fitting in multi dimensional hyperspectral imaging datasets, being used by 100+ members of my international team.
• Completed advanced training in CUDA programming, developing custom GPU kernels for tasks in both astrophysics and quantitative finance (e.g., finite difference schemes).
Python Deep Learning Analyse et prévision des séries chronologiques Data science IA et analyse de données
STATISTICS WITHOUT BORDERS
PROJECT & CLIENT MANAGER (PRO-BONO)
ENERGIE
avril 2024 - Aujourd'hui (2 ans et 2 mois)
London, Royaume-Uni
• Led a team of five data scientists and researchers to model energy systems and assess decarbonisation strategies for European markets using Python and open-source datasets for an Italian climate think tank.
• Responsible for project scoping, volunteer recruitment, stakeholder coordination, and final delivery, including documentation and knowledge transfer.
Python Développement et évaluation de modèles IA et analyse de données Project Management Team management
UNIVERSITY COLLEGE LONDON
POSTDOCTORAL RESEARCHER
CENTRES DE RECHERCHE
mars 2020 - août 2022 (2 ans et 5 mois)
London, Royaume-Uni
• Built pipelines for analysing time-series and image datasets from the Very Large Telescope, including noise modelling, source detection, model fitting and computer vision for image identification/classification.
• Employed Bayesian modelling and MCMC for parameter inference and uncertainty estimation in astrophysical models to interpret observed, multimodal spectroscopic/time-series data.
Machine learning Data science Python Modélisation prédictive Analyse et prévision des séries chronologiques