We build a space infrastructure as a service, providing rapid, reliable, and simplified access to space.

Part of the Ground Software team, I am developing Cockpit, Loft's satellite operations software that empowers end-users with powerful satellite constellation management capabilities while abstracting away the complexity of spacecraft operations.

I am also supporting satellite commissioning and daily operations as a Flight director, contributing to the success of Loft’s SatDevOps movement, a highly automated and efficient way to operate satellites.

Development of an AI research assistant to keep up with the topics, trends and industries you care about, without the information overload.

Meet Leo: https://blog.feedly.com/leo/

As first Machine Learning Engineer at GlobalEnglish, I developed from scratch two significant machine learning products:

Automatic English Lesson Generator:

● Designed, in collaboration with linguistic experts, a Machine Learning software to generate English learning lessons automatically

● Researched and implemented two recent papers on Neural Question Generation using PyTorch to create advanced Reading Comprehension questions automatically

● Deployed the project as a Flask microservice

Automatic English Pronunciation Scorer:

● Developed deep learning algorithms to score English pronunciation automatically after researching and experimenting state-of-the-art architectures mixing CNN and LSTM models

● Built a real-time dashboard to monitor the model performance over time and avoid concept drifting

● Investigated Speaker Verification methods including D-vector extraction to prevent fraud during English Speaking tests

● Leveraged the Speech Recognition engine Kaldi for Speech-to-Text and Speech Alignment tasks

● Worked on speech transcripts of non-native English speakers to score Grammar skills automatically using statistical NLP and deep learning models implemented in PyTorch

Consulting mission with Accenture aimed at developing Accor Hotels' loyalty program using past transactional data:

● Identify Accor Hotels' best members using k-means FM segmentations

● Qualify best members travelling behaviors

● Target best future members using scoring models

● Devise a personalized acquisition campaign strategy to recruit these future members of interest

● Deliver recommendations to a C-level audience

fifty-five helps brands leverage their data to increase marketing, advertising, and sales effectiveness.

● Analyse data and structure relevant information for retail clients, in consistency with business indicators specific to their industry (Google Analytics, Adobe Analytics)

● Build reporting tools monitoring client’s digital activity in real-time, based on key performance indicators (Tableau Software)

● Advise a French bank on a set of tools to use for their banking app in order to monitor customers' activity, create marketing campaigns (in-app message or push notification) and fix crashes