About Aziz
Senior GenAI Engineer with a strong focus on building production-grade AI systems for enterprise environments. Currently contributing to the AI platform of Canal+, one of the world's leading media and entertainment groups, where I design and ship end-to-end Generative AI solutions used at scale.
My core expertise spans AWS Bedrock (Agents, Knowledge Bases, Prompt Management, Guardrails), RAG architectures, hybrid search (BM25 + vector), LLM orchestration with LangChain/LangGraph, and recommendation systems built with AWS Personalize, LightFM and Recombee — all deployed on a modern AWS serverless stack.
My approach to AI is grounded in a rigorous mathematical and scientific background — holding degrees from Paris-Dauphine (AI & Data Science), and strong foundations in Applied Mathematics, Stochastic Analysis, and Econometrics. This depth allows me to go beyond tooling and reason at the system and model level.
French
Native or bilingual
English
Native or bilingual
Remote only
Primarily works remotely
Experience
- Canal +AI ConsultantAugust 2024 - Today (1 year and 10 months)GENERATIVE AI & LLM APPLICATIONSDesign and development of LLM-powered assistants, AI copilots, and enterprise knowledge platformsImplementation of RAG architectures integrating enterprise data sources via Bedrock Knowledge Bases and OpenSearchDevelopment of enterprise Knowledge Hubs enabling intelligent access to internal documentation, analytics datasets, and operational knowledgeIntegration of LLM tool usage and orchestration pipelines using Bedrock Agents, LangChain and LangGraphPrompt lifecycle management and versioning using Bedrock Prompt ManagementImplementation of AI safety and content moderation layers via Bedrock GuardrailsModel evaluation and benchmarking using Bedrock Model Evaluation for quality assurance of LLM outputsImplementation of prompt orchestration pipelines and LLM experimentation frameworksAWS AI SERVICESEnterprise search and knowledge retrieval using Amazon Kendra for intelligent document indexing and Q&AConversational AI interfaces built with Amazon Lex for intent recognition and dialogue managementNLP pipelines leveraging AWS Comprehend for entity recognition, sentiment analysis, and text classificationMultilingual AI systems using AWS Translate for cross-language content processing and localizationML model training, deployment and monitoring pipelines using Amazon SageMakerAI SEARCH & RETRIEVAL ENGINEERINGDesign of hybrid search architectures combining lexical and semantic retrievalImplementation of BM25 lexical search for precise keyword-based document retrievalIntegration of vector search pipelines enabling semantic discovery of enterprise knowledgeDevelopment of reranking layers to improve retrieval quality before LLM generationOptimization of retrieval quality and relevance scoring for enterprise AI assistantsAI ARCHITECTURE & ORCHESTRATIONDesign of AI system architectures integrating LLMs, search engines, APIs, and data platformsImplementation of LLM orchestration pipelines using LangChain and LangGraph.
- IFP Energies NouvellesIngénieur en IAFebruary 2022 - July 2024 (2 years and 5 months)Rueil-Malmaison, France- NLP & LARGE LANGUAGE MODELSAdvanced NLP acceleration on Tesla supercomputer (720 nodes × 8 NVIDIA A100 GPUs) for large-scale text analysis (APPIS Project)Developed sophisticated NLP systems integrating Retrieval-Augmented Generation (RAG) and fine-tuning methodologiesLed LLM + Knowledge Graph integration to significantly boost contextual understanding of complex entities in scientific textsPushed NLP frontiers with transfer learning and adaptive learning rate techniques for unprecedented text analysis accuracy- DEEP LEARNING & BIOINFORMATICSLed codon optimization algorithm refinement in the Digital Lab, incorporating insights from landmark biology research tailored for internal datasetsUtilized Topaze supercomputer (AMD EPYC Milan 7763) with efficient parallel GPU processing for large-scale model trainingPioneered next-generation codon optimization methods via advanced deep learning algorithms, improving genetic code translation efficiencyMentored internship project focused on 3D rock imagery analysis using deep learning- COMPUTER VISION & LARGE VISION MODELSEnhanced Rocknet rock type classification models using Topaze Nvidia A100 processors for high-resolution geological imagery processingEnhanced Large Vision Models (LVMs) with novel neural network architectures and optimization techniques for improved image recognitionApplied cutting-edge training and refinement strategies on complex visual tasks across scientific datasets
- INRIAIngénieur de Recherche en Traitement de Données LIDARMay 2021 - December 2021 (7 months)Palaiseau, France- MACHINE LEARNING & PREDICTIVE MODELLINGImplemented and benchmarked predictive ML models: Random Forest, MLP, XGBoost with rigorous cross-validation (LOOCV, K-Fold, Stratified CV)Designed supervised models for LIDAR bathymetric data using annotated resources from SHOMDesigned unsupervised models for coastal data with significantly different characteristics from the annotated training setCreated new sparse databases to support model generalization on novel coastal profiles- SIGNAL & IMAGE PROCESSINGComparative analysis on spatial regularization filters: Mean, Median, Gaussian, Bilateral, Bilateral GradientComparative analysis on signal denoising filters: LowPass, HighPass, BandPass, BandStop, Kalman, Extended Kalman, Wiener, HilbertModeled and analyzed filter performance across multiple LIDAR signal conditions- DEEP LEARNING & RESEARCH IMPLEMENTATIONTraining and validation via Transfer Learning and state-of-the-art U-Net architectures on Inria's GPU clusterImplemented from scratch (no existing code): BathyNet, DANAE++ and other state-of-the-art bathymetric modelsHyperparameter optimization and statistical result presentation (curves, confidence intervals, input/output diffs)Delivered proof-of-concept to SHOM for integration into operational systems
Recommendations
Be the first to recommend Aziz
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Master en Intelligence Artificielle et Sciences des Données, Intelligence artificielle et Sciences des DonnéesUniversité Paris Dauphine2020Master en Intelligence Artificielle et Sciences des Données, Intelligence artificielle et Sciences des Données
- Master1 Méthodes Quantitatives , Economie Mathématique et EconométrieUniversité de Tunis El Manar2019Master1 Méthodes Quantitatives , Economie Mathématique et Econométrie