Media content

↓ Scroll to Explore

ABOUT

Building Cloud, AI, and Scalable Systems

I'm a Master's student in Computer Engineering at UTD, focused on ML-powered and automated systems. I architect production-grade cloud infrastructure, build full-stack applications, and design interfaces that turn complex systems into something people can actually use.

AWSKubernetesTerraformPythonReactFlutterDockerCI/CD

CAPABILITIES

Full-stack. Cloud-native. Production-ready.

Cloud Architecture

AWS, Terraform, K8s, CI/CD pipelines, and observability at scale.

Product Engineering

Full-stack systems that turn complex workflows into usable interfaces.

Backend & APIs

Service design, data modeling, and integrations across modern stacks.

Execution Quality

Refactoring discipline and code that stays maintainable after launch.

SELECTED WORK

Projects built for production.

EvalForge — Multi-Dimensional AI Eval Harness
AI & EVAL

EvalForge — Multi-Dimensional AI Eval Harness

Standalone harness that scores any LLM/agent on RAG groundedness, tool-use reliability, and guardrail robustness — with versioned YAML suites, a CI regression gate, and a live dashboard.

PythonFastAPIRAGCI/CDRender
RadiTriage — AI Radiology Workflow Assistant
AI & HEALTHCARE

RadiTriage — AI Radiology Workflow Assistant

Agentic veterinary radiology pipeline: classify urgency → RAG over prior cases → draft a preliminary report → route to a specialist queue. Role-based access, SLA ops dashboard, and key-safe Claude proxy with demo fallback.

ReactViteNodeClaudeRAGRBAC
Agentic AI Governance Harness
AI & GOVERNANCE

Agentic AI Governance Harness

Reproducible benchmark harness for agentic AI governance — measures guardrail trip rate, policy latency, intervention frequency, and cost per resolved task.

PythonLLM JudgeBenchmarksRender
Nexus SRE — Cloud Cost & Reliability Agent
AI & CLOUD OPS

Nexus SRE — Cloud Cost & Reliability Agent

Autonomous FinOps + SRE agent: Bedrock reasoning over CloudWatch, Cost Explorer, and Trusted Advisor — with human-in-the-loop approval before any write remediation.

BedrockFastAPILambdaCloudWatchCDK
FieldTech Assistant
FULL-STACK / AI

FieldTech Assistant

RAG-powered assistant for field technicians with manuals, wiring context, service history, and offline-first PWA.

FastAPIReactChromaDBRAGPWA
ML Experiment Tracker MCP
ML & TOOLING

ML Experiment Tracker MCP

MCP server for experiment runs, metrics, hyperparameters, and tags with SQLite persistence and a live dashboard.

PythonMCPSQLiteStarlette
AWS CloudOps MCP
MCP & CLOUD

AWS CloudOps MCP

Read-focused MCP server for AWS ops with a live dashboard and synthetic inventory for demos.

PythonMCPAWSboto3
Small Business Operations MCP
MCP & OPS

Small Business Operations MCP

MCP server for tasks, expenses, CRM, and inventory with finance CSV analytics and a live ops dashboard.

PythonMCPSQLiteStarlette
Budget Insights Platform
CLOUD & FINANCE

Budget Insights Platform

Cloud-first finance platform with Flask microservices, React dashboard, and anomaly detection.

ReactFlaskDockerAWS
Geofences Platform
PRODUCT

Geofences Platform

Geofencing platform with Apple-inspired UI, real-time notifications, and policy automation.

MapsReal-timeNotificationsAutomation
pgvector RAG Demo
ML & DATA

pgvector RAG Demo

RAG chat app with PostgreSQL pgvector retrieval, embeddings, and Streamlit UI with citations.

PythonPostgreSQLpgvectorStreamlit

Let's build something together.

I'm open to full-time roles, contract work, and collaborations in cloud engineering and product development.

© 2026 Oscar Valles