Naif ASSWIEL
Machine Learning Student
About Me
Machine Learning Student
I'm a Machine Learning student with a strong software foundation. I have a Bachelor's degree in Computer Science, and I am currently pursuing a Master's in Artificial Intelligence at Le Mans University, where I focus on Natural Language Processing and Automatic Speech Recognition, studying both foundational concepts and state-of-the-art Deep learning models. I have experience in AI research and am eager to contribute to real-world applications.
Work Experience
Intern @ LIUM
- Applied and fine-tuned state-of-the-art models for audio-visual Deepfake detection.
- Reproduced state-of-the-art results through large-scale model training.
- Leveraged temporal Deepfake detection models to run experiments and assess their performance on benchmark datasets.
Publications
Making Hallucinations Useful: A Reassessment of the Troublemaker Agent Strategy in the Age of Generative AI
I contributed to two research papers on the same topic, one published in English and the other in French, presented at two conferences, including one international The topic proposes a platform designed for medical students, using a RAG-based system to generate yes/no affirmation-style questions. The system retrieves validated information from a knowledge base and deliberately introduces controlled inaccuracies to stimulate critical thinking. This approach explores how hallucinations can be transformed into a pedagogical tool.
My Role:
- Built the RAG pipeline and evaluated it with different LLMs and SLMs.
- Assessed the quality of question generation and retrieval using targeted evaluation metrics.
Education
Master of Artificial Intelligence
Le Mans University
Focusing on Machine Learning algorithms, Deep learning, Natural Language Processing, Signal Processing, Data modeling, Neural Networks, and Experimental evaluation of AI systems.
Bachelor of Computer Science
University of Montpellier
Focused on software development and programming, algorithms and logical reasoning, data structures, databases, and computer networks.
DELF B2 - French Language
CAVILAM – Alliance Française
Learned French from scratch through intensive classes, interactive projects, and cultural activities.
Projects
Improving Pronunciation in Conditional Flow Matching TTS Models (ongoing)
- Set up and traine a Conditional Flow Matching TTS system using the Blizzard Challenge 2023 dataset.
- Traine a speech recognition model (Wav2Vec-BERT) for objective evaluation of TTS output.
- Analyze the impact of text-acoustic alignment on pronunciation accuracy.
- Implemente reinforcement learning methods to reduce transcription errors and improve model robustness.
TrAP - TroubleMaker Agent Platform
- Developped a pedagogical agent for medical education using LLMs and SLMs for generate complex questions.
- Implemented a retrieval-augmented generation (RAG) mechanism to ensure all questions are backed by existing resources from the medical database.
- Designed the frontend using React for building dynamic user interfaces and Django for efficient API handling.
Credits to: Idir Akretche, Lucas Schmitt and Amine Kessemtini
AnimusVision.
- Implemented CNNs architectures for binary and multiple classfication.
- Applied transfer learning and optimization techniques to reduce overfitting and improve accuracy.
- Implemented GAN architecture for image genration
- Implemented a Variational Autoencoder (VAE) using latent space representations to colorize, denoise, and generate images.
Real-Time Car Detection System
- Integrated YOLOv3 neural network with OpenCV DNN module for real-time vehicle detection.
- Implemented video processing pipeline with preprocessing, detection, and post-processing stages.
- Applied Non-Maximum Suppression (NMS) algorithm to filter overlapping detections and improve accuracy.
Intelligent PDF Question-Answering System with RAG
- Built a Retrieval-Augmented Generation (RAG) system that processes PDF documents and generates semantic embeddings using HuggingFace transformers.
- Integrated Google Gemini LLM for intelligent question-answering based on document content.
- Implemented vector indexing and similarity search to retrieve relevant context and generate accurate responses.
Deep Dive AI (ongoing)
This project provides a set of notebooks designed for hands-on practice with core Machine-learning concepts.
- Linear Regression - House Price Prediction.
- Logistic Regression - Titanic Survival Prediction
- Residual connections with CNNs.
Full E-commerce Backend with Node.js and RESTful API.
Developed a comprehensive e-commerce backend system from the ground up, covering everything from basic CRUD operations to advanced features. This project served as a practical Deep-dive into backend development, helping me understand how complex business logic comes together in production-ready applications.
- Implemented full CRUD functionality for the platform’s main entities.
- Handled file uploads with image processing
- Developed authentication and authorization system with JWT tokens, including password reset functionality and role-based access control.
- Designed modular architecture with separate models, services, routes, and middleware for scalability.
- Integrated payment processing system for transaction handling.
ShopEase
- Develop a comprehensive e-commerce mobile app using Flutter for the front-end and Dart for framework support.
- Implement a FastAPI and Firebase backend to manage data, authentication, and server-side operations.
Credits to: Amine Kessemtini
Certifications
Deep Learning Specialization Certificate
- Explore core neural network architectures such as CNN, RNN, LSTM, and GRU.
- Use transfer learning by building on architectures like ResNet and MobileNet.
- Gain hands-on experience with Deep vision applications such as image segmentation and object detection.
- Study attention mechanisms and transformers, and their applications in tasks such as Machine translation and NER.
Image Segmentation with PyTorch Certificate
- Hands-on practice for image segmentation using U-Net.
Machine Learning Specialization Certificate
- Gained hands-on experience with the fundamentals of Machine learning, including key algorithms.
- Practiced tasks such as classification, linear regression, and clustering to apply these concepts.
AI Foundations for Everyone Certificate
- Familiarized with key AI concepts and terms, including AI, ML, and DL.
- Learned about different types of AI (generative and discriminative) and their real-life applications.
Skills
Machine Learning and Data Analysis
TensorFlow, Transformers, PyTorch, Scikit-Learn, NumPy, Pandas
Programming Languages
Python, SQL, C, JavaScript
Frameworks & Tools
Git, SLURM, MongoDB, Node.js/Express.js, FastAPI, Django, Ollama.
Soft Skills
Communication, Problem-solving, Adaptability, Time Management, Technical Writing.
Languages
- French: Fluent
- English: Fluent
- Arabic: Native
Volunteering
Humanitarian Association of Montpellier Certificate
- Distribution of meals to people in precarious situations.
- Teamwork to carry out the mission while respecting deadlines.
Get In Touch
I'm always open to exploring new opportunities, working on interesting projects, or exchanging thoughts on AI and broader aspects of life.