Mauricio González Orti

I build production AI systems spanning industrial computer vision, LLM agents, data pipelines, and backend infrastructure. My physics and mathematics background helps me turn complex technical problems into reliable deployed solutions.

At Parvada, I built an end-to-end Python and C++ AI pipeline that automated chicken classification at 3,000 chicks per minute. I also reduced inference latency by 33%, improved model accuracy by 60%, and owned deployment, hardware integration, validation, and monitoring across three live production systems.

At Okus AI, I designed an ELT pipeline that doubled production throughput by converting factory video into structured operational records using CNN pose detection and segmentation. I also built DynamoDB and Amazon Bedrock infrastructure that helped double the number of clients.

I continue building tools such as ScienceCompanion, a multi-agent RAG system using LangChain, LangGraph, OpenSearch, and LLMs, and MauriGPT, a custom and fine-tuned LLM deployed with FastAPI and Streamlit. I enjoy improving system efficiency and collaborating with clients and stakeholders to deliver impactful AI infrastructure.