Mohammed Khairallah — AI / ML Engineer

Mohammed Khairallah — AI / ML EngineerSkip to main content BasedParis, FranceRoleAI / ML engineerFocusAgentic AI · ArchitectureAvailableJun 2026 MohammedKhairallah I architect Agentic AI systems and ML infrastructure for enterprise workflows. From high-performance FastAPI microservices to containerized model pipelines—I design the seam where intelligent systems meet production reality. Book a call Selected work Hire on Upwork Available for projects --- About, briefly. Profile I build Agentic AI systems and the architecture that lets them run in production—FastAPI services, vector stores, model-serving pipelines, and the orchestration layer that ties them together. My training at 42 Paris was systems-first: C, networking, virtualization. That foundation shows up every time I have to make a model serve under real latency, real load, real cost. I work best when the problem is concrete—extracting structure from messy industrial data, replacing brittle workflows with agentic pipelines, or making a slow inference path five times faster. School42 Paris Years4+ writing software StackPython / TS / C++ MethodShip → measure → revise LanguagesEN · FR · AR --- Selected work. Agentic ML platform — orchestration for enterprise workflows Multi-agent system on FastAPI + LangGraph: routes industrial requests across specialized agents, with vector retrieval, tool use, and observable traces.FastAPILangGraphpgvectorRedisDocker2026AI ArchitectRepogithub ML model-serving stack — containerized inference at scale Production pipeline that packages PyTorch + scikit models behind a typed FastAPI gateway, with autoscaling workers and GPU-aware batching.PyTorchFastAPIDockerTritonK8s2025ML EngineerRepogithub Document intelligence — structured data from messy PDFs RAG pipeline with layout-aware OCR, hybrid retrieval (BM25 + dense), and an LLM extractor that emits validated JSON for downstream systems.PythonLlamaIndexPostgresTesseractOpenAI2025Full-stack