I am a senior R&D engineer and backend developer with a special interest on Machine Learning, Semantic Web, Big Data, Information Retrieval and Functional Programming.
I have joinded KCityLabs (Kinaxia Group) in August 2017. I am both contributing to development and research projects. First, as a Backend Developer, I am in charge of the development of an API that controls the whole life cycle of a geographical knowledge base, which aggregates more than 20 data sources. Then, as a Data Scientist, I am conducting a research project, in collaboration with CNRS, to predict dwelling prices. In particular, we are training a machine learning regressor able to handle different types of noises (e.g. ontic imprecision, missing data, outliers).
I have worked 3 years at Vigiglobe, which terminated its activity in June 2017. I contributed to the development of a social media analytics platform which was mining and aggregating thousands of messages per second in real time. We fully developed this application with functional programming languages (Clojure / Haskell) and NoSQL frameworks including Druid, Elasticsearch, Kafka and MongoDB. I have in particular developed components for brand recognition, geolocation, textual search, and statistical trend detection.
I previously worked in various types of organizations: research institute, big corporations and startups. I have started with a Ph.D. thesis at Orange Labs (in collaboration with INRIA and Telecom ParisTech) on semantic social network analysis. My results were published in proceedings of both national and international conferences, including the International Semantic Web Conference. In 2011, I joined Orange’s search engine team where I build a knowledge base supplying both internal business intelligence tools and search engine components. This knowledge base was aggregating heterogeneous data from both internal sources and external ones and was taking advantage of Semantic Web technologies to reifer identities, align models and infer/recover data. This knowledge base was in particular used to detect named entities (ex: brands, persons, places) in query logs to help marketing team understanding customer needs. Before joining Vigiglobe, I have worked one year in a first startup, Mnemotix, as an R&D engineer and Semantic Web developer. My main contribution was the development of a semantic layer to link autonomous medical databases for conducting large-scale epidemiologic analysis.