faycel amrouche

Designed agentic workflows for cloud-edge infrastructure using LLM-driven decision-making, enabling autonomous network management with reduced human intervention. Built dynamic orchestration logic for self-healing networks that monitor traffic, predict congestion, and reallocate compute resources in real time.

Investigated memory management protocols for LLM agents, comparing episodic vs. semantic memory to improve long-term reasoning. Optimized episodic knowledge retrieval to support consistent, context-aware agent performance.

Led end-to-end development of a Django backend to automate validation pipelines for industrial sensor streams, migrating a 15-day manual process to real-time execution under 1 hour. Enforced code quality standards and statistical integrity testing to ensure 100% data integrity for global production models.

Architected scalable Flask APIs to serve RAG-based AI assistants for mission-critical drilling operation data. Improved response times using asynchronous processing and optimized SQL/vector search queries, and built reusable retrieval templates to accelerate internal AI initiatives.

Re-engineered cross-validation protocols to prevent data leakage, improving the robustness of downstream models. Delivered measurable gains in model performance by strengthening evaluation integrity.

Designed a mission-critical licensing platform for the Ministry of Agriculture using Django and PostgreSQL. Built a secure administrative dashboard that streamlined national maritime logistics for thousands of users.

Activities and societies: National Excellence: ranked 1st in the Province in Baccalaureate (18.84/20); top 25% of the first historic promotion of the school.

State Engineer Degree program in Artificial Intelligence (Master’s equivalent) at ENSIA, completed from 2021 to June 2026. Achieved top academic standing, including ranking 1st in the province in the baccalaureate.