Skander Ghariani

I’m a Software Engineer focused on backend and applied AI. At CURE (Berlin-based, remote), I built a production pipeline around a locally-hosted, quantized Mistral 7B model (llama.cpp) to parse German pharmacy call transcripts into structured order data.

I developed CureSearch, a product search engine with embeddings-based recommendations and therapeutic (ATC) classification (Node.js, MongoDB). I also integrated Uber Eats, Wolt, and Lieferando for order fulfillment, and created admin dashboards and internal tools using Angular and TypeScript.

Before that, I researched parameter-efficient fine-tuning methods (LoRA, Delta-LoRA) for LLM optimization as a Research Intern, explored event-driven microservices and Kafka architecture, and built an AI Assistant desktop application with a local LLAMA/Alpaca model (Flask backend, VueJS frontend). I’m available for fully remote roles in the CET timezone.