Alice Coucke
Research Scientist at Sonos Inc
Paris, Ile-de-France
Summary
👤
Senior🎓
Top SchoolAlice Coucke is a research scientist and machine learning leader with a decade of experience building privacy-preserving, embedded speech and language systems. She has led ASR and ML research teams at Sonos and contributed foundational work at Snips—shipping on-device voice control to millions while publishing and patenting advances in keyword spotting, federated learning, and edge NLU. Her portfolio spans research, productization and large-scale evaluation with a strong focus on fairness, privacy and efficiency, including an open dataset and bias-assessment methodology published at LREC-COLING and a special session she organized at Interspeech. A trained physicist (PhD, ENS), she blends rigorous quantitative methods with hands-on engineering—contributing to notable open-source projects like snips-nlu and quality-testing in Rasa. Known as a frequent conference speaker and industry spokesperson, she also mentors teams and shaped hiring and interview practices to grow ML talent.
10 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy (PhD) Statistical Physics, Doctor of Philosophy (PhD) Statistical Physics at Ecole normale supérieureClasses préparatoires, Classes préparatoires at Lycée Henri IVMaster 2 ENS ICFP Theoretical Physics, Master 2 ENS ICFP Theoretical Physics at Ecole Normale Supérieure
French, English, German
Github Skills (19)
pytest10
python10
feature-extraction10
scikit10
nlu10
machine-learning10
scikit-learn10
nlp10
test-automation10
rasa9
chatbot9
data-augmentation8
text-classification8
entity-extraction8
named-entity-recognition8
Programming languages (4)
Github contributions (5)
snipsco/snips-nlu
Mar 2017 - Apr 2017
Snips Python library to extract meaning from text
Role in this project:
ML Engineer
ML EngineerContributions:80 commits in 17 days
Contributions summary:Alice primarily contributed to the development of an intent classifier within the Snips NLU library. Their work included implementing feature extraction techniques using CountVectorizer and TfidfTransformer, along with feature selection methods. They also designed and implemented data augmentation strategies and integrated the intent classifier with the Snips NLU framework. The user focused on model training, and improving the intent classification accuracy through parameter tuning and model selection.
nlupythonbotintent-classificationtext-processing
RasaHQ/rasa
Feb 2017 - Feb 2017
💬 Open source machine learning framework to automate text- and voice-based conversations: NLU, dialogue management, connect to Slack, Facebook, and more - Create chatbots and voice assistants
Role in this project:QA Engineer / Test Automation Engineer
QA Engineer / Test Automation EngineerContributions:5 commits, 1 PR, 1 comment in 1 day
Contributions summary:Alice primarily contributed to the testing framework of the Rasa project. Their work focused on adding and modifying tests, specifically for the tokenizers, which are critical for Natural Language Understanding (NLU). The commits involve creating new test cases, ensuring the correct tokenization of various input strings, including those with special characters and offsets. This indicates a focus on quality assurance and ensuring the reliability of the NLU components within the chatbot framework.
nlupythonbotspeech-recognitionbotkit
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.