Senior Software Engineer | GenAI (AWS) | AI Platform & Infrastructure Engineering

I specialize in building secure, deterministic, and cost-efficient AI infrastructure on AWS.
My focus is on reproducible pipelines, GPU FinOps, Terraform governance, and secure RAG architectures. Tools that make AI systems safer, cheaper, and easier to operate in real production environments.

Design philosophy: deterministic outputs, security-by-default, offline execution, zero-IAM assumptions.

I build open-source scanners that surface high-impact risks in AI and GPU workloads -before they reach production.

High-signal analysis for vector databases and RAG pipelines.
Detects unsafe retrieval patterns, embedding leakage, and configuration drift.
👉 https://github.com/Dee66/VectorScan

GPU cost-governance and utilization analysis.
Identifies cost drift, underutilization, scaling anomalies, and waste patterns.
👉 https://github.com/Dee66/ComputeScan

These modules provide enforceable policies and exportable audit results for teams that need stronger governance signals.
All scanners remain fully open source.

Zero-Trust RAG governance with reproducible OPA policies and audit-ready compliance outputs.

GPU FinOps governance with utilization insights, drift analysis, and reproducible cost reports.

A set of real Terraform plans demonstrating vector security flaws, GPU drift, and misconfiguration scenarios validated by GuardSuite scanners.

AWS-native reference architecture demonstrating secure GenAI patterns, reproducible MLOps pipelines, and cost-aware infrastructure.
👉 https://github.com/ShieldCraft-AI/ShieldCraft-AI

Minimal AWS-native framework for RAG pipelines, PEFT fine-tuning workflows, and secure MLOps deployment.
👉 https://github.com/Dee66/CodeCraft-AI

• Deterministic engineering & reproducible pipelines
• Terraform plan parsing & governance
• OPA/Rego policy engines
• Vector DB & RAG security
• GPU FinOps, rightsizing, and drift detection
• AWS-native AI (Bedrock, SageMaker, Lambda, CDK)
• Secure MLOps: CI/CD, tagging, environment-aware configs
• Offline-first, audit-friendly infrastructure tooling

• Expanding ComputeScan with GPU drift autopsy reports
• Adding new RAG security checks to VectorScan
• Growing the GuardSuite blueprint ecosystem for broader AI governance

Email: support@shieldcraft-ai.com
LinkedIn: https://www.linkedin.com/in/dee-prinsloo
Website (coming soon): shieldcraft-ai.com