I build things at the intersection of AI and data — LLM-powered agents, automated reporting pipelines, and dashboards that actually get used. Over the past 3+ years I've shipped production solutions for clients across the EU, USA, and MENA using LangChain, Azure OpenAI, Power BI, and SQL.
My graduation research took LLMs beyond chat — using them as an evolutionary engine to autonomously design metaheuristic algorithms, cutting the optimality gap from 23% down to 5.5% on classical NP-hard benchmarks. Currently working on LLM agent infrastructure and WMS data systems.
Divide and Conquer & Total Conquer: Architectural Evolution of Metaheuristics using LLMs Graduation thesis · Tanta University · 2025
Used LLMs as an evolutionary engine to autonomously construct metaheuristic architectures — extending ReEvo. TC strategy co-designed a hybrid Memetic Algorithm from a standard ACO with zero manual intervention.
5.50% optimality gapvs23.25% baseline·p < 0.001· TSPLIB · ACO · TSP · CVRP