🌐 Portfolio: mohamed-ashraf-ai.vercel.app
I'm an AI Engineer specializing in Generative AI, LLMs, RAG systems, and autonomous AI agents — with hands-on experience shipping production-grade AI to real users, not just notebooks. I build multi-agent systems with Google ADK and LangGraph, fine-tune LLMs and VLMs for specific domains, and care most about the part that makes AI actually reliable: grounding, guardrails, and evaluation. I also work deeply in Arabic NLP and Computer Vision, building AI that genuinely works in Arabic — not just English.
- RAG Systems: Hybrid retrieval (dense + keyword), reranking, and grounded generation — evaluated with RAGAS, not guesswork.
- AI Agents: Multi-agent orchestration with Google ADK and LangGraph — tool use, guardrails, memory for production chatbots.
- LLM & VLM Fine-Tuning: LoRA/QLoRA/GRPO for domain-specific models — including building datasets from scratch when none exist.
- Full-Stack AI Deployment: FastAPI, Docker, PostgreSQL, Qdrant/pgvector — from prototype to crash-safe, monitored production systems.
LANGUAGES
AI & DATA SCIENCE
GEN AI & NLP
BACKEND & DEPLOYMENT
