Julián C.

• • Generative AI Strategy and Architecture (LLM & RAG): I designed the implementation of the corporate AI ecosystem (Vision + Generative), building advanced RAG architectures for real-time material verification and validation. I created the end-to-end pipeline by automating the generation of multimodal embeddings with CLIP to align visual and text data, their indexing in vector databases, and hallucination mitigation, ensuring maximum technical accuracy and strict compliance with healthcare data governance.

• • Productization and Deployment (MLOps & Cloud): I conceptualized and developed the conversational voice interface for healthcare personnel, inspired by efficient intelligent assistant architectures (Amazon Alexa style). To this end, I integrated and optimized local Small Language Models (SLMs) and those from the Gemma family, enabling linguistic processing at the edge (local) to issue and receive verbal instructions with critical latency, without the need for physical contact in the operating room.

• • Inference Optimization and High Availability: I created the complete model lifecycle (CI/CD, Docker, monitoring) under scalable microservices and RESTful API architectures. I successfully led the system's transition from the pilot phase to its final productive implementation in hospital environments and Cloud infrastructure.

• • Data Engineering and Automation (Computer Vision & NLP): I led the deployment of computer vision (YOLO, PaddleOCR) and language models in critical environments, optimizing memory and computation usage to ensure ultra-low response times (inference <400ms in vision). I automated key operational workflows for ingesting and digitizing complex documentation and implant cards with over 98% accuracy.

• • Large-scale Digitization: I led the massive implementation of EDI systems connecting over 50 public hospitals, automating critical processes and reducing processing times by 60%.