Luis San Martin

I am Luis, a Chilean-Valencian LLM AI Technology Architect with 10+ years in software engineering and 6+ years in AI. I specialize in AI architecture and LLMOps/AgentOps for enterprise systems, leading on-premise and hybrid deployments aligned with ENS, GDPR, and the AI Act. I design and evaluate LLM/SLM, computer vision, and AI agent platforms for client projects and EU tenders.

Principal

Lead of I+D Commission in the Chilean Chamber of Artificial Intelligence (CCHIA).

AI/LLM Technology Architecture Assoc. Manager at Accenture (2026-03 – Present)

Technical Lead in AgentOps and AI Architecture for a flagship initiative focused on agentifying enterprise automations across hybrid environments (cloud and on-premise). Responsible for designing and delivering a scalable, container-based AI platform leveraging open-source technologies for secure and sovereign deployments.

• Architected and deployed agent-based automation systems across hybrid (cloud/on-prem) environments
• Defined AgentOps architecture, including SDD, guardrails, evaluation frameworks, and multi-model orchestration
• AgentOps support to evaluate ML and generative AI models using XAI metrics
• Research, documentation, and development of POCs - R&D
• Successfully transitioned AI workloads from cloud-only to hybrid/on-prem architectures, enabling data sovereignty and cost optimization
• Implemented observability and evaluation pipelines for GenAI systems, improving reliability and reducing production risk

MLOps/LLMOps Lead Engineer at Exceltic (2024-03 – 2026-03)

Technical Lead in MLOps/LLMOps and AI Architecture, driving production-grade computer vision systems for real-time video analytics and scalable generative AI platforms with strong governance and lifecycle management.

• Development of MLOps architecture using MLFlow and DVC
• Implementation of CI/CD processes with Docker and Kubernetes in the cloud and on-prem
• QAOps support to evaluate ML and generative AI models using metrics
• Research, documentation, and development of POCs - R&D
• Improved CI/CD infrastructure to optimize the deployment of YOLO CV models for various use cases
• Developed the MLOps architecture required to obtain and analyze performance metrics for the different models and their versions in production using tracking, versioning, and monitoring

AI Engineer at Mercedes Benz Group (2023-09 – 2024-03)

Development of AI/ML-based web applications to optimize internal company processes based on predictive AI models.

• IT infrastructure support and model deployment to Azure Cloud Databricks
• Development of an end-to-end product that provides access to the model through a REST API with FastAPI
• Implementation of CI/CD pipelines with Docker and Azure
• Research and technical documentation
• Deployment of AI models in production with integration into enterprise infrastructure
• Process optimization with deployment of solutions in Azure Databricks and real-time monitoring

AI Engineer at Amaris (2021-12 – 2023-09)

Participation in European-funded R&D projects for the development and implementation of ML and Generative AI-based solutions for the massive generation of synthetic data and the development of proprietary products.

• Building AI/ML models
• Generating synthetic datasets for model training
• Developing the model inference backend service using a REST API with FastAPI
• Implementing CI/CD pipelines with Docker and Jenkins
• Building an asynchronous messaging service using Kafka
• Web scraping with Python, Selenium, and BeautifulSoup
• Research and technical documentation
• Creation of efficient pipelines for training and deploying AI models
• Development of synthetic data generation techniques to improve model performance
• Implementation of graph-based search algorithms for semantic search in real-time web scraping

Duale FP Degree in Computer Science and Software Development – HHS Karlsruhe (2019 – 2021)