Esmeralda C. Cabrera Ventura
Cloud · Software · Applied AI

Architecture decisions, trade-offs, and what they cost in production.

I'm Esmeralda — a Solutions Architect, U.S. Air Force veteran, and founder of Apex Intelligent Systems. My portfolio shows what I've shipped; this is where I document how it was engineered and why.

About

My path into engineering ran through the U.S. Air Force, and the habits it built — plan before you execute, think in systems, own what you ship — still shape how I approach technical work. I hold a Master of Science in Computer Science, maintain a security clearance, and work as a Solutions Architect and the founder of Apex Intelligent Systems.

Most of what I do lives between architecture and implementation: standing up production cloud environments across AWS, Azure and hybrid infrastructure, building cloud-native applications, and developing machine learning and agentic AI automation. Cloud engineering, software development, DevOps and applied ML usually arrive as one problem rather than four, so I work across all of them — building secure, scalable, high-availability systems for government and commercial environments, with an eye toward operational resilience and modernization that produces measurable outcomes.

Away from the keyboard, I put my energy into fitness, continuous learning, and giving back through STEM education and veteran causes. Curiosity is the thread running through most of it.

Writing is how I think a problem through properly. Each article here takes one system I've worked on and walks the reasoning: the constraints, the trade-offs, what a decision cost, and what I'd approach differently next time.

Portfolio and resume  ·  LinkedIn  ·  GitHub

Articles →

July 2026 · 9 min read

Agents as services, not prompts: engineering a deterministic multi-agent SOC analyst

Five specialist agents, each deployed as its own addressable inference endpoint on IBM watsonx.ai, chained under a supervising orchestrator — and the decision to trade open-ended agent autonomy for a deterministic single-pass pipeline.

July 2026 · 9 min read

Proof of identity, not proof of humanity: a serverless gate for a public calendar

Engineering an identity-gated scheduling system on AWS Lambda, API Gateway, S3 and CloudFront — with a three-state policy engine whose admission rules are configuration and whose verdicts are proven across the entire input space.

July 2026 · 8 min read

Inside Fridge2Meal: a deterministic recipe engine built to reduce food waste

Why I engineered the core matching intelligence directly in PostgreSQL and Node.js, and designed the generative AI layer to sit beside it as an optional enhancement rather than a dependency.

Project detail

Agentic AI · Engineering dossier

Autonomous Cybersecurity Multi-Agent System

Six deployed AI services on IBM watsonx. Five specialist agents chained under a supervising orchestrator, grounded by trusted hybrid retrieval, executed as a deterministic single pass.

Serverless · Engineering dossier

Recruiter Verification Gateway

A serverless identity gate on AWS. Six services, one public route, a five-test tri-state policy engine proven across all 162 of its states, and signed fifteen-minute approval credentials.

Full-stack · Engineering dossier

Fridge2Meal

An ingredient-first meal platform. Five orchestrated services, 16 normalized tables, weighted compatibility ranking, Keycloak identity, and a containerized runtime that starts from one command.