December 2022 - Present

DataBron
Stockholm, Sweden

CEO and Senior Data Scientist

I am developing the overall vision and strategic direction my consulting company while driving its growth and success.

• I am working with international clients to provide data-driven solutions to their issues in analytic and modelisation.

March 2020 - April 2023

Data Analytic Team, Tradera
Stockholm, Sweden

Lead Data Scientist

I was leading the data science vision and roadmap for the e-commerce platform Tradera; while deploying machine learning solutions to improve circular consumption and sustainable shopping. Proficient in handling the whole data lifecycle from warehousing to deployment and conveying the outcomes intelligibly to boost the core business.

• Implemented recommendation systems for personalised brands, search keywords, and sellers, and trending searches. Increased number of clicks on these features by, at least, 400\%.
• Developed tags auto-completion by NLP approaches, dynamic auto-modeling for each attribute, 91 models in production.
• Completed two user safety projects on phishing monitoring and frauds detection based on five years of data, >5B observations.
• Designed four A/B tests and their statistical validations based on Bayesian hypothesis. At least 2\% improvement on core metrics under 5\% MDE.
• Promulgated data science within the company through three workshops and two lectures, one being at the G-Cloud Day.
• Defined and supervised two Master student projects, on the category tree architecture and deep learning implementation of product recommendations.

October 2019 - Present

Forest Phenotyping Working Group
International Plant Phenotyping Network

Chairman

Organising the development of the Forest Phenotyping Working Group. The goals of the Forest Phenotyping Working Group is to improve the understanding of multiple scale forest dynamics through its phenotypes. The development of tools, platforms, methodologies and an international community will bring together experts from silvilculture, forest management, genetic, computer & data science, experimental design, and forest biometrics and geomatics, to tackle the emerging challenges related to climate changes, sustainable forest management, the need for healthier forests and the manufacture of sustainable and cleaner products.

• Promote the concept of forest phenotyping to the international scientific community.
• Enable the sharing of knowledge and efforts with the international scientific community
• Encourage the expansion of efforts to phenotype seedlings, trees, and forests
• Build networks within the IPPN community to share and develop platforms of interest to other IPPN working groups

October 2019 - February 2020

Data Analytic Team, Scion
Rotorua, New Zealand

Research Leader

Supporting Digital Foresters in a data-driven World.

• I was leading the Data Analytics Team comprised of three data/computer scientists, one bio-informatician and one biometrician, working across the forestry value chain to optimise its sustainability and the transition to a bio-circular world.

January 2018 - October 2019

Data Analytic Team, Scion
Rotorua, New Zealand

Data Scientist

Phenotyping Platform

• Implemented gradient boosting models to handle large, forest-scale dataset with complex and noisy features with mixed data types for phenotyping inferences. These inferences allow to estimate and predict the productivity of trees across the estate, and therefore to select tree stocks based on environmental conditions for superior genotyping.
• Improved and maintained a web platform for visualisation of large point-cloud data. I am currently Project Leader on the internal deployment of this platform.
• Supported the statistical analysis of several business-oriented projects, including client reporting.

January 2016 - January 2018

Centre for Image Analysis and SciLifeLab, Uppsala Universitet
Uppsala, Sweden

PostDoc

Large-Scale Data Analysis for Digital Image Analysis Applications I developed a tool for automated quantification of cell and tissue morphology in digital images. Difficulties lie in the parallel management and processing of 24 images (e.g., 48000x40000 pixels). This framework aims to support clinicians' diagnosis and to complement visual assessment when investigating disease and/or drug response.

• Developed an open-source tool which combines the analysis of gene expression with quantification of cell and tissue morphology.
• Shaped a web platform for visualisation of large slide scanner images (e.g., 48000x40000 pixels) at different resolutions.
• Developed an image processing algorithm for nucleus and lipid droplet segmentation and feature extraction in high-content/high-throughput microscopy screening.

October 2012 - October 2015

Université Blaise Pascal
Clermont-Ferrand, France

PhD Student

Characterisation of Volcanic Emissions through Thermal Vision

I studied the different components of strombolian eruptions at the full range of remote sensing spatial scales. These range from millimeters (83,000 individual particles detected through 31 eruptions recorded at 200Hz with thermal camera) to kilometers (two frames per day over two years, imaged with satellite field of view). Overall, I provided a better understanding of plume dynamics through thermal vision.

• Developed an algorithm to segment and track high-speed particles recorded on thermal videos.
• Designed a novel method to segment and parameterise volcanic plumes on thermal videos.
• Implemented an algorithm to detect multiple change points in 2D radiometer data.
• Investigated new processes to detect hot spots in satellite imagery.

April 2012 - September 2012

Lawrence Berkeley National Laboratory (LBNL), Berkeley, USA

Research Trainee

Java applications development for Detection and Tracking of DNA repair centres

• The aim of this project was to improve the understanding of the nucleid organisation and the spatial distribution of the DNA repair centres. Due to the large amount of data involved, I translated the existing codes in Java. Then, I developed an algorithm to segment, track and register nucleid through time.

April 2011 - September 2011

Laboratoire des Sciences de l'Habitat
Lyon, France

Intern

iPhone application development for lighting design

• I developed an iPhone application to model luminance maps in interior spaces: iPhotoLux. I carried out several experiments to calibrate the camera of the iPhone4. Then, I updated an application which provides daily brightness data of a position to optimise photovoltaic panels unfolding.