Marouan B.

I’m an AI & Software Engineer with 2+ years of experience designing and building production-grade AI systems, with a focus on LLMs, RAG, agentic workflows, and intelligent data pipelines. My work covers the full AI engineering lifecycle — from experimentation and architecture to backend development, evaluation, deployment, and optimization. I’ve worked on systems involving: → LLM & RAG Applications — Building production AI assistants with advanced retrieval, hybrid search, reranking, and multi-step reasoning. → Agentic AI Systems — Designing workflows with routing, tool use, decision graphs, fallback mechanisms, and self-reflection to handle complex tasks reliably. → Document Intelligence — Transforming complex PDFs and unstructured documents into validated, structured data through automated extraction and processing pipelines. → LLM Optimization — Improving inference and serving performance, including reducing LLM response latency by approximately 80% through optimized model serving. → Production AI Infrastructure — Deploying and operating AI services using Docker, Kubernetes, GCP, APIs, orchestration pipelines, monitoring, and scalable backend architectures. My core stack includes Python, LangGraph, LangChain, LangSmith, vLLM, Qdrant, FastAPI, Dagster, PostgreSQL, Docker, Kubernetes, and GCP, alongside the broader LLM and open-source AI ecosystem. What interests me most is the engineering behind AI: turning models into systems that are accurate, observable, scalable, and useful in real-world environments. Always interested in challenging AI engineering problems, especially around LLMs, RAG, Agentic AI, AI infrastructure, and production AI systems.