Nouha B.

– Built automated data pipelines with multimodal embeddings and graph ontology.

– Enhanced open-set object detection through prompt automation and improved accuracy in previously unseen object classes, with high-precision labeling.

– Implemented multi-modal RAG framework with graph ontology integration.

– Keywords: Multimodal Models, Image-Text Embeddings, Open-set Object Detection, Object Segmentation, Elasticsearch, Graph Ontology, GraphDB, Dimensionality Reduction, Text-to-Text from Images, FastAPI, PyTorch, KNN, CNN, Data pipelines, Microservices, MultiModal RAG, SSH, Linux

– Proposed and developed an agentic recommendation system to optimize test and validation efforts, and reduce execution time. The system was designed as an (MVP) with a strong focus on scalability, security measures and compliance.

– Defined agent goals, tasks, and tools for modularity and scalability.

– Built a Batch on Demand ETL pipelines from large, complex, and diverse data sources, and using NLP techniques.

– Researched RAG, fine-tuning, and function-calling capabilities, selecting the optimal method for each use case.

– Experimented with open-source LLMs, evaluated their features, and selected the suitable one for the requirements.

– Quantized LLMs, tuned parameters, and applied prompt engineering.

– Integrated advanced technologies in GenAI based on scientific papers and continuous monitoring of innovations.

– Co-supervised 3 interns, and presented weekly progress updates to the internship jury members.

– Keywords: Agentic system, PyTorch, NLP, Data pipelines, Data engineering, ETL, LLM, Jira, GitLab, XRay

– Designed and Developed a web application for influence marketing. – Worked on the project specifications and implemented REST APIs, authorization, and authentication using JWT, Spring Security, Spring Boot, and Linux OS. – Used Instagram API to get the influencer's information and posts in real-time. – Created the frontend side using React and Redux.