ENGINEERING_MANIFEST.yaml

The SRE Workbench & Architecture Log

> SYSTEM.STACK_UPDATED: 2026-05

Engineering is the art of managing trade-offs. This is my curated stack—optimized for high-velocity Kubernetes operations, deep-layer observability, and agentic AI infrastructure.

01 Cloud-Native Platforms

ADR Control Plane

Amazon EKS

The foundation of my distributed systems. Migrating control plane overhead to AWS allows me to focus on cluster scaling, cgroup isolation, and raw compute efficiency.

Multi-AZ / Karpenter Scaling
ADR IaC / Provisioning

Terraform

Infrastructure must be immutable. I rely on declarative HCL state management to enforce consistency and prevent configuration drift across multi-AZ environments.

Zero-Drift Enforcement
ADR GitOps

ArgoCD

Eliminated `kubectl apply` entirely. ArgoCD acts as my automated reconciliation loop, guaranteeing cluster state strictly mirrors the Git repository source of truth.

Automated Reconciliation

02 Telemetry & Observability

METRICS

Prometheus & Grafana

My primary engine for SLI/SLO governance. I map custom dashboards to the Four Golden Signals to instantly identify latency spikes and infrastructure saturation.

Golden Signal Mapping
LOGGING

ELK Stack / Loki

Centralized log aggregation is non-negotiable for rapid RCA. Loki integrates perfectly with Grafana for seamless correlation between metrics and trace logs.

Log Correlation
SECURITY

Trivy & OPA Gatekeeper

Shift-left security. Gatekeeper enforces strict admission controller policies, preventing non-compliant or vulnerable container images from entering the cluster.

Admission Control

03 Agentic AI Infrastructure

INFERENCE

Amazon Bedrock

My primary gateway for orchestrating LLM inference and deploying highly available agentic workflows without managing underlying GPU clusters.

Serverless LLM Ops
INTEROP

Model Context Protocol (MCP)

The standard for interoperability. I use MCP to standardize AI context integration across models, enabling secure and predictable agentic infrastructure.

Standardized Context

04 Engineering Workbench

COMPUTE

Dell G15 (Intel i7-HX)

High-performance x86 compute environment. I leverage WSL2 for deep-level native Linux kernel access, utilizing the raw multi-core HX architecture to run local multi-node Kubernetes topologies (Kind) and demanding infrastructure simulations.

WSL2 / Linux Kernel
SHELL

WezTerm + Fish Shell

A GPU-accelerated terminal emulator paired with Fish's intelligent auto-completion. This combination provides the raw speed necessary for rapid infrastructure navigation.

GPU-Accelerated UX
IDE

Neovim (LazyVim)

My primary IDE for Infrastructure as Code. Optimized for keyboard-centric development with high-fidelity LSP support for Go, Python, and Terraform.

Keyboard-Centric Dev