Freelancer profile translated to English.

Description

AI Engineer specialized in Artificial Intelligence (Machine Learning) with a strong knowledge of software engineering. I have extensive experience in using Deep Learning methods applied to text and voice data, as well as in using Large Language Models (LLMs). I also have significant experience in integrating AI solutions into new or existing environments, covering the entire lifecycle, from study to evaluation and deployment.

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 20km), Toulouse (up to 10km)

Lokela
AI Lead / Software Engineer
EDUCATION AND E-LEARNING
October 2022 - Today (3 years and 8 months)
Toulouse, France
Startup aimed at making language learning easier and more fun for children, by leveraging the latest innovations in generative AI. The application offers the following features:
● Creation of tailor-made stories enriched with narration, adapted to each user's preferences.
● Ability to answer questions and generate quizzes based on the chosen story.
● Control of correct pronunciation and tracking of memorization.
Technical Environment: Python, Javascript, React, Azure AI, Firebase storage, firebase serverless functions, NLP: ChatGPT,
LLama, Mistral, prompt engineering, Image: Midjourney, Stable diffusion, runawayML, Speech: Whisper...
BANQUE DE FRANCE
Senior Data Scientist / Software Engineer
BANKING AND INSURANCE
March 2021 - June 2022 (1 year and 3 months)
● Needs analysis for speech recognition for an application, evaluation and implementation of a benchmark of several paid and open-source solutions.
● Development of the backend in Python/Django, deployment in the form of Docker containers.
● Adaptation of a speech recognition model (NLP) to best meet the vocabulary of the application in question.
● Taking over an NLP project on user question answering / Chat:
● Evaluation of several technical approaches that match the team's resources.
● Rasa model configuration/training.
● Elasticsearch configuration.
● Evaluation of Neural Embeddings (semantic search) for query/response similarity.
● Integration of new components into the main application, testing, etc.
Technical Environment: Python, DOCKER, wav2vec, custom LSTM Speech model, Pyctcdecode, KenLm, Hugging Face, Azure
AI, Rasa, Tensorflow, Spacy, Django, Elastic search, neural embeddings (NLP vector search)
LINAGORA
Machine Learning Engineer / Data Scientist
SOFTWARE PUBLISHING
January 2018 - November 2019 (1 year and 10 months)
● Collection/cleaning of a large volume of data (10,000 hours of voice/text data with annotation) to train Deep Learning models.
● Development and testing of several architectures and optimization algorithms in Pytorch to enable rapid model training (CNN, LSTM, TRANSFORMERS, attention/CTC).
● Development of a compact, stateless REST API to facilitate scaling.
● Docker packaging, Kubernetes, and deployment on Managed Kubernetes instances in the Cloud.
Technical Environment: Python, Pytorch, DOCKER, Kubernetes, TorchServe, Spacy, KenLm, OVH CLOUD