Abdelilah Elgallati

I've built and evaluated end-to-end AI systems spanning multimodal CTR prediction, RAG pipelines, recommendation engines, and AI microservices.

At eGlobal Digital, I contributed to IVEP, a production-grade multi-role SaaS platform with 30+ backend modules and AI services. I designed a hybrid recommendation pipeline and built multilingual RAG conversational services using Sentence-Transformers, LangChain, ChromaDB, Whisper, and NLLB-200.

For the WWW 2025 MM-CTR Challenge, I developed a cascaded multimodal recommendation pipeline using 3.6M training samples and achieved a 0.9495 validation AUC with a Transformer, Deep Cross Network, and DNN.

I've also built full-stack products including MyInvoice, a platform for SMEs and freelancers, and deployed RAG and recommendation applications with FastAPI, Flask, React, MongoDB, and Docker.