NLP Engineer
Feb. 2018 - present
Digimind, Grenoble, France
- Design, conduct, and develop company's first automated short text (tweet) classification model in the multilingual context. The micro-service built upon this model is capable of classifying real-time streaming data from Kafka.
- Employ state-of-the-art deep learning tweet sentiment analysis model, which outperforms the baseline by at least 20% absolute improvement on english and french.
- Design, conduct and develop a machine learning service detecting tweet users’ location
- Contribute to design a logo detection model using transfer learning along with a scene text recognizer. In production, this model achieves about 90% on both precision and recall.
- Build several internal tools to speed up model serving process such as a Flask bootstrap service and an out-of-the-box monitoring service
Machine Learning Developer (part-time)
Oct. 2017 - Jan. 2018
BingBin, Compiègne, France
- Employ transfer learning using Mobilenet for garbage image classification. The model is used in a smart trash bin that recognises the trash automatically.
- Build an online annotation system to collect our own dataset of images of trash.
Machine Learning Research Internship
Feb. - Jul. 2017
Amadeus, Nice, France
- Research on natural language processing with deep learning.
- Implement diverse neural networks on Tensorflow to find best contact points from a given text. The NLP engine's accuracy has been improved from 22% to 67%.
- Deploy the microservices for the model and create new API.
Project Developer
Dec. 2016 - May. 2017
USEC, Junior Entreprise de l’UTC
- Build a NoSQL database to store client's test data.
- Write a french manual for the database and script usage.
Python Developer Internship
Sep. 2015 - Feb. 2016
Lengow, Nantes, France
- Develop E-commerces’ connectors using Python.
- Create and update Lengow’s APIs on Django framework.
- Bug fix on JIRA.
IT Project Engineer Internship
Apr. - Aug. 2014
Oxylane Group (Decathlon), Shanghai, China
- Test and compare different equipments in terms of network performance and stability.
- Help project manager make decisions.
- Contact and negotiate with equipment's provider.
- Update team’s site.