Hello! 👋
I'm Adam MOUNIR
Research Student majoring in Data & Artificial Intelligence at EFREI Paris. Passionate about AI research, deep learning and building intelligent systems.
About Me
Research student at EFREI Paris Panthéon-Assas, specializing in Data & Artificial Intelligence. My research focuses on deep learning generalization, explainable AI, and bio-inspired neural architectures. I have hands-on experience in LLM fine-tuning, NLP pipelines, and causal reasoning — with internships at Thales and Société Générale.
// languages
// technical skills
Projects
🧠 Double Descent & Loss Landscapes →
Investigating the generalization paradox of over-parameterized DNNs. Analyzing SGD implicit bias, minima flatness, and the Double Descent phenomenon using Sharpness-Aware Minimization and 2D loss surface visualization.
🔬 Neuro-AI for Complementary Learning Systems →
Tackling catastrophic forgetting via bio-inspired architectures optimizing the stability-plasticity trade-off. Built a Wake-Sleep/DDPM framework for NREM/REM latent generative replay, surpassing VAE/GAN inference speed & retention.
💡 EXPLAiN — XAI for Patent Classification →
Built explainable AI models (Transformers/Shallow) for patent classification, fine-tuning BERT and FastText. Integrated Shapley Values for interpretability and deployed a web interface to visualize feature contributions. Achieved 75% F1-score.
Experience
AI Research Intern
Thales — Paris
Researching global-scale Root Cause Analysis by distilling multi-step reasoning from LLMs into specialized architectures. Designed a supervised alignment pipeline mapping expert causal trajectories from logs into internalized reasoning schemas. Distilled causal traces via GPT-4o to fine-tune a SFT pipeline using Llama 3.1 (QLoRA). Achieved 0.64 F1-score in RCA prediction.
AI Engineer — Apprentice
Société Générale — Paris
Detecting semantic discrepancies between unstructured document disclosures and structured multi-provider data. Engineering a Zero-shot NER pipeline using Claude 3.5 Sonnet (RAG) and Directed Graphs for topological impact analysis. Automating cross-source data alignment and causal influence tracing within hierarchical scoring models.