D

deep-learning

The AI and the Automobile collection brings together multiple specifications that define how AI integrates into modern vehicle systems, treating the automobile as a software-defined platform where intelligence spans perception, control, navigation, and energy management. It emphasizes real-time performance, safety-critical architecture, and the transition from purely mechanical systems to continuously evolving software-driven mobility. Across the collection, the specs also describe how automotive AI should be built through open and interoperable systems with strong safety practices, including simulation, validation pipelines, and human-in-the-loop oversight to handle edge cases. Together, they frame vehicle intelligence as a layered system combining autonomy, redundancy, diagnostics, and secure update mechanisms to enable reliable and scalable deployment.

Photo-realistic single-image super-resolution (SRGAN, x4) in TensorFlow/TensorLayer — refactored OO training/evaluation pipeline with research experiment branches.

IF-Net fork + complete preprocessed ShapeNet archive (tar.xz bundles via git LFS). ShapeNet derivatives: non-commercial research use only.

PyTorch reimplementation of CheXNet: multi-label classification and CAM/Grad-CAM localization of 14 thoracic diseases on ChestX-ray14, with full-resolution (1024px) ResNet50 training.

Academic NLP project focused on multilingual text summarization using Transformer-based deep learning models and natural language processing techniques.

Intelligent VRAM/RAM swapping for LLM inference - Extension of KVortex | Offloading intelligent VRAM/RAM pour l'inference

https://ayinedjimi-consultants.fr

Automated LLM Benchmarking on GPU - tokens/sec, latency percentiles, VRAM profiling, multi-format support (HuggingFace, GGUF, GPTQ)

https://ayinedjimi-consultants.fr

VRAM to RAM Offloader for AI and vLLM - High-Performance C++23 KV Cache Engine with Multi-Stream GPU Transfers

https://ayinedjimi-consultants.fr

LLM quantization & benchmarking on GPU - GGUF, GPTQ, AWQ, bitsandbytes | Quantification et benchmark de modeles LLM sur GPU

https://ayinedjimi-consultants.fr

GPU-accelerated embedding server for RAG systems - CUDA, FastAPI, sentence-transformers | Serveur d'embeddings GPU ultra-rapide

https://ayinedjimi-consultants.fr

Deep learning system for seismic velocity inversion using a SincNet-GAT-UNet architecture, physics-informed temporal encoding, graph attention fusion, and FiLM-conditioned spatial prediction

Introduction to classification using machine learning and deep learning (PyTorch, TensorFlow, Keras)

Projet pédagogique en algorithmie et intelligence artificielle consistant à créer un agent (Kevin) capable de résoudre un labyrinthe.

Le projet explore plusieurs approches :

Génération de labyrinthes (DFS, Prim) Algorithmes de recherche de chemin (A*, Dijkstra) Apprentissage supervisé (Imitation Learning avec CNN) Apprentissage par renforcement (Deep Q-Network)

Des outils de visualisation permettent de générer des images et des GIFs montrant Kevin se déplacer dans le labyrinthe.

Projet réalisé dans le cadre de la formation Développeur en Intelligence Artificielle (Simplon).

对不同尺寸的交通信号灯图片做红黄绿分类

A clean, modular implementation of Physics-Informed Neural Networks (PINNs) for solving the 1D and 2D wave equation using PyTorch.

LeNet-5 in 9 lines of code using Keras, pour la lecture d'étiquettes marquées à la main provenant de la caméra du robot.

This is a passion project where I attempt to use deep-learning ML-models to make accurate predictions on the outcome of NHL games. I ran the model predictions as a live betting strategy in the Fall of 2022 and achieved a roughly 60% return on investment.