Freelancer profile translated to English.

Description

👋 Hello! Engineer specialized in Natural Language Processing (NLP), I am passionate about artificial intelligence and always ready to put my technical skills to the service of stimulating and varied projects. My expertise in NLP has been developed through various concrete achievements, and I am available for any mission requiring expertise in this field.

🌟 Key Skills:

NLP 📝:

- Development of machine translation tools

- Sentiment analysis

- Data extraction and implementation of question-answering systems on textual documents

- Design and development of chatbots

- Automation using large language models (LLM)

- LLM Fine-tuning

- Advanced text vectorization techniques

- Automatic content generation

- Text classification and categorization

Web Development 🌐:

- 🔧 **Back-end & API**: Design and implementation of back-ends and APIs for your applications

- 🎨 **UI/UX**: Development of user interfaces for optimized interaction with AI models

🚀 Recent Achievements:

- 🤖 Design/Development of a chatbot specialized in the legal field

- 🎯 Fine-tuning of language models for specific tasks

- 📄 Development of a tool for automatic information extraction from PDF documents

If my profile matches your needs or if you would like to know more about my skills and experience, do not hesitate to contact me. I would be delighted to discuss how I can contribute to the success of your projects.

Looking forward to hearing from you! 🤝

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 10km)

Legomnia
Development of an advanced search platform for legal documents
LEGAL
September 2024 - November 2024 (2 months)
- Development of a platform using ElasticSearch 8.9, with integration of document backup in vector format and a hybrid retriever, combining syntactic and semantic searches for increased efficiency in exploring complex legal corpora.
- Implementation of advanced queries, allowing for relevant and targeted information retrieval through algorithms optimized for the specificities of legal documents.
- Automatic extraction and indexing of metadata, offering better filterability and intuitive document management.
- Complete management of the API and backend, including the design, development, and maintenance of backend services, ensuring seamless integration with ElasticSearch and robust performance.
Python Document search Data Extraction NLP Elasticsearch Web development
Renault
NLP Engineering Intern
AUTOMOBILE
September 2022 - September 2024 (2 years)
Paris, France
Fine-tuning of the Mistral 7B LLM model:
- Fine-tuning of a large language model (LLM) to perform specific tasks with innovative approaches such as Quantization and QLoRA, significantly reducing resource requirements while maintaining optimal performance.
- In-depth analysis of performance and adjustment of hyperparameters to ensure increased consistency and accuracy in results.

Technologies used: Python, PyTorch, Transformers, QLoRA, Fine-Tuning, Quantization.

Development of internal chatbots:
- Creation of search solutions based on the Retrieval-Augmented Generation (RAG) framework, combining an existing ElasticSearch database and vector databases to meet contextual extraction needs within internal documentation bases.
- Integration of a hybrid retriever combining semantic and syntactic search to maximize result relevance.
- Full development: FastAPI for the backend API and integration with internal systems.

Technologies used: ElasticSearch, LLM, RAG, Python, FastAPI.

Structured information extraction:
- Design of an automated extraction pipeline based on models like LayoutLM and BERT, with OCR integration to process scanned documents.
- Structuring of extracted data for use in analytical pipelines.
3/5

Technologies used: Python, Transformers, LayoutLM, OCR, BERT, PyTorch.
NLP Pytorch TensorFlow Machine learning Deep Learning artificial intelligence LLM Fine-tuning OCR Transformers Elasticsearch Retrieval Augmented Generation RAG
Vinymatic
Image recognition tool
ARTS AND CRAFTS
February 2023 - April 2023 (2 months)
Toulouse, France
- Development of a high-performance algorithm based on SIFT and FAISS to quickly identify images with 95% accuracy and an inference time of 0.03 seconds.
- Optimization of the infrastructure to ensure scalability in production environments.

Technologies used: Python, OpenCV, FAISS, SIFT.
Computer Vision artificial intelligence