Salah-Eddine I.

• Architected and implemented a redesign of the SHARP processing chain, reducing computation time by over 40% through JIT compilation with Numba, parallelization, and memory I/O optimization.

• Redefined the software architecture for offline operation, adapting computation chains and integrating robust switching mechanisms to ensure operational continuity.

• Designed and deployed in production an anomaly detection and correction module for critical time series. This innovation was the subject of a patent application, securing the company's intellectual property.

• Developed and trained Machine Learning models for sensor data reconstruction, enabling expanded analysis capabilities to non-equipped helicopter variants.

• Developed an end-to-end software decoder in Python for a new avionics unit (ARINC 429 protocol), from technical specification to production deployment.

• Developed and deployed a chatbot using AI agents to connect to technical databases, enabling support teams to efficiently answer complex technical questions through an intuitive conversational interface.

• Migrated the software infrastructure and data pipelines to AWS, modernizing legacy architectures to cloud-native solutions to ensure high availability and dynamic scalability.

Technologies used: Python, Shell, Bash, Unittest, Pytest, Pandas, Numpy, Numba, Cython, Gitlab, Docker, ARINC 429, Elasticsearch, PyTorch, TensorFlow/Keras, Scikit-learn, RAG Architecture, LLMs (OpenAI API / Mistral), LangChain / LlamaIndex, Vector databases (ChromaDB / Pinecone), AWS (S3, EC2, ECS, Lambda, IAM).

• Contributed to the development of a web application in micro-services architecture (Java, Angular) and implemented the CI/CD deployment pipeline with Docker, ensuring reliable production releases.

• Automated non-regression tests (Selenium, TestNG) for the Sun Set application, improving deployment reliability and providing detailed test reports.

• Developed and deployed a Deep Learning model (CNN) for diffraction data analysis, accessible to researchers via a dedicated web interface.

• Implemented a solution for anonymizing Oracle databases using SQL scripts and a web interface, ensuring GDPR compliance for test environments.

Technologies used: Angular, Java, Docker, Selenium, Oracle DB, SQL.

Mentor for Morocco AI Generative AI Hackathons : InovAI2024, GENAI2023