marcosrevilla.dev

SYSTEM_INIT

Hi. I'm Marcos.

I'm a Telecommunications Engineer with a professional background in systems engineering, currently focused on AI and Generative AI.

Alongside my professional work, I build end-to-end AI systems — including multimodal agents, RAG pipelines and Python backend services — with an emphasis on architecture, evaluation and deployment.

I also care about how technology is experienced and presented, with a particular interest in product design and creative direction.

Featured work includes VeriRAG, NeuroVid Analytics, Spotibind and Corrientes MET — spanning AI systems, data-driven applications and full-stack product engineering.

Professionally, I currently work as a Security Engineer at Capgemini.

Explore the projects, capabilities or career history — or ask me anything about them.

Core Capabilities

Stack Overview

> CORE_CAPABILITIES

01_AI

AI & Agentic Engineering

Bounded tool-using agents, observable RAG systems, and GPU inference pipelines.

Core Tech: LangChain, LangGraph, MCP (Model Context Protocol), Local & Cloud LLMs (Ollama, Claude, GPT-4o), ChromaDB.
Applied Experience: NeuroVid is a bounded multimodal agent with a real MCP client/server boundary: stdio sessions, tool discovery and three read-only tools over a frozen analysis snapshot. Anthropic/OpenAI adapters stay separate from that tool layer. VeriRAG covers inspectable RAG, grounded answer policies and retrieval evaluation. GPU inference includes Meta TRIBE v2 on RunPod. Structured outputs and human review reduce unsupported claims; they are not treated as proof.

02_DATA

Data Engineering & Backend

Designing asynchronous, high-throughput pipelines and real-time APIs.

Core Tech: Python, FastAPI, AsyncIO, Pandas (ETL), Server-Sent Events (SSE), MySQL, MongoDB.
Applied Experience: Orchestrating parallel cloud/local workloads to collapse latency. Experience building real-time data streaming architectures, custom rate limiters (Upstash Redis), and robust relational state management for massive datasets.

03_STACK

Full Stack & Systems Architecture

Shipping production-grade, secure, and fully observable web platforms.

Core Tech: Next.js (React 19, TypeScript), Tailwind CSS 4, Auth.js (Edge Middleware), RunPod, Vercel.
Applied Experience: Deploying complete end-to-end architectures—from GPU microservices and strict Zod API validation to state-machine-driven CMS platforms and real-time collaborative environments.