Florent Hippolyte Heuberger Email & Phone Number

France

Lyon, Auvergne-Rhône-Alpes, France

Coordinating a team of 4 ML Engineers around the productisation of Target Identification and Drug Repurposing pipelines. The product is two-fold: 1. A dashboard of Machine Learning results aggregated from patient data, clinical trials, public biology datases and LLM summaries used by Biomedical Researchers as a basis for Drug Discovery decisions, 2. Libraries used by Data Scientists to compute and enrich the results with additional patient data information and to adapt to different therapeutic areasOur team focuses on the development of these libraries built around efficiency and modularity.

Lyon, Auvergne-Rhône-Alpes, France

Created a team of 3 ML Engineer dedicated to bringing software tools to 90 ML Research Scientists, Computational Biologists and Biostatisticians. Our responsibilities are the following:* Automate Target Identification and Drug Repurposing pipelines based on patient data, clinical trials, public biology databases and LLM extractions* Maintain several python libraries dedicated to Data Science at Owkin* Infuse software engineering practices across Data Science activity including support to research projects and pharma projects* Serve as a bridge between Data Science and Software Engineering activities at Owkin (through management of change, and DS needs aggregation)

Paris, Ile-De-France, France

Developed and structured a team of 8 Data Scientists and 2 Machine Learning Engineers around a scalable and reproducible histology image analysis pipeline.We automatized the creation of AI applications combining traditional image analysis and recent deep learning techniques. The AI applications are all built from a monorepo focusing on the following:* Creating an image quantification baseline as a one-line command* Bringing flexibility to conduct research on any components of the pipeline* Setting up a framework for collaborative and traceable experiments (through the use of cloud providers)The company and the products were then acquired by Tribun Health at then end of 2022.

Paris Area, France

Keeneye offers a platform for storing, sharing, visualizing and analyzing medical images. It supports any microscopy image format, and allows navigating in the slides as well as annotating them. Everything is in the browser thus no software needs to be installed and the work can be easily shared through a simple link.Furthermore our platform makes it possible for the user to create an application that automatically detects and segments all objects similar to their annotations.Ready applications, able to detect predefined cells/objects are also at the customer’s disposal. As part of the R&D team, my role is as follows: - Deep learning research applied to microscopy images, - Putting in production the machine learning models that have been developed within the team, - Monitoring of scientific publications related to deep learning & computer vision.

Copenhagen Area, Denmark

Combining Deep Learning and Spectral Graph theory, Spektral specializes in real-time video segmentation. I am currently working on Multi-task deep learning for keypoint localization. I work on implementing a multi-task architecture for simultaneous segmentation and keypoint localization, with state-of-art results on the MSCOCO and MPII keypoints.

Copenhagen Area, Denmark

Canecto is building a tool intending at giving recommendations to webmasters on how to improve their website. These recommendations include visualisation dashboards allowing the customers to better understand his users. Our tool makes it possible to analyse the interactions of users with the text in the website. For instance the impact of a couple of significant keywords in the web pages or the influence of level of language are estimated.I was Data Scientist and Django Developer to implement the back-end of the python based web application. I was especially working on deploying and optimizing the speed of simple NLP models among a team of 4 developers / data scientists.

London, Royaume-Uni

Solocal UK is the European leader in online advertising and performs targeted display for more than 20 000 customers. I took part in the creation of a bidder for Real Time Bidding advertising:- Prediction of the probability of click using a machine learning model : Field Aware Factorization Machine.- As part of a team of five developpers, realized an algorithm to select within 20 ms a sample of campaigns that would be relevant for a user.

Paris

As the Quantitative Analyst department of Deloitte, the team takes part in different Risk Management projects as well as model reviews. I worked on two projects :- Developping a java tool that implements the Basel comitee publication IRRBB (Interest Rate Risk in the Banking Book) of June 2015,- As part of the 2016 stress tests, evaluating the Net Interest Income and the credit risk provision.Both projects involved valuating most vanilla financial products as well as banking retail products.

Master'S Degree, Mathematical Modeling And Computation

Applied Mathematics on Machine Learning. My first four years of studies have had a strong focus on Mathematics and Computer Science. My.

Master, Applied Mathematics

Classes Préparatoires Aux Grandes Écoles Pcsi - Psi*