Reymark Garcia

Designed and led the development of a multi-tenant SaaS platform delivering customizable LLM/RAG-driven services, fully branded per customer via isolated tenant subdomains. Implemented a production-scale multi-agent AI system using LangChain, LangGraph, OpenAI APIs, and Model Context Protocol to enable standardized, extensible AI tool orchestration and context sharing across agents and services. Implemented MCP-based integrations to expose internal services (RFP parsers, document stores, scoring engines, retrieval APIs) as reusable, discoverable tools for LLM agents, enabling clean separation between model reasoning and platform capabilities. Implemented Human-in-the-Loop workflows with approval gates, audit trails, override mechanisms, and LLM-as-a-Judge evaluation layers integrated into MCP-enabled agent flows for automated quality control, compliance checks, and decision validation. Optimized AI and platform performance through prompt engineering, vector index tuning, document chunk

Designed and led the development of a multi-tenant SaaS platform delivering customizable LLM/RAG-driven services. Architected end-to-end backend and frontend components, including FastAPI microservices, Next.js front end with role-based access and dynamic tenant UIs, a PostgreSQL database with pgvector, and a containerized cloud deployment. Implemented RAG workflows for RFQ/RFQ responses and built production-grade ML pipelines, monitoring, and security practices.

Built an enterprise-grade, production AI performance-monitoring chatbot to automate large-scale data retrieval and accelerate real-time troubleshooting for support engineers in high-throughput environments. Designed and implemented a production medical imaging pipeline for echocardiogram analysis using PyTorch, Lightning, MONAI, and OpenCV, delivering automated labeling, classification, and grading. Developed an AI-driven automation platform for an aseismic construction company, generating project documentation and compliance reports. Built React-based dashboards integrated with Elasticsearch to improve reporting accuracy, observability, and operational efficiency. Built and deployed predictive ML models for a SaaS platform using PyTorch, scikit-learn, XGBoost, and MLflow, integrating inference into MERN-based dashboards to surface predictive insights directly within core product workflows.