BatistaCloud — Junior Batista · Cloud & Network Engineering

00The Plan The idea driving the whole project.

The goal: a personal AI assistant that works like an employee — and stays private.

I want an assistant that can control my home, send me updates, run tasks, and hold my personal context — but built on a principle most cloud products ignore: my data stays on my own server, in my home. The heavy reasoning runs through cloud LLM APIs, but the context, memory, and control all live locally on hardware I own.

Instead of renting someone else's cloud, I'm building my own — BatistaCloud — and using it as a real-world classroom for the cloud & network engineering degree I'm working toward. Every piece I add is a skill I'm learning by doing: virtualization, containers, networking, VPNs, secure remote access, and AI integration.

Private by design

Personal data and memory stay local. Cloud APIs do the thinking, never the storing.

Self-hosted

Runs on my own hardware under Proxmox — no rented servers, full control of the stack.

Learn by building

Every feature is a networking or cloud concept put into practice, documented as I go.

01The Stack Live status of each system. Updated as the build evolves.

Proxmox VE 9

Bare-metal hypervisor on the home server. Runs every VM and container — the foundation the whole platform sits on.

live

Home Assistant OS

Deployed as a KVM virtual machine. The control plane for the smart home — lights, climate, and devices.

live

Private AI Agent

Self-hosted assistant in its own LXC container, running as a persistent service. Reachable over Telegram; reasoning powered by Claude with a fallback model.

live

Agent → Home Assistant

The AI agent is wired into Home Assistant with a scoped access token, so it can read state and control real devices.

live

Tailscale Mesh VPN

WireGuard-based private network with the server as an exit node — secure remote access from anywhere, no public ports.

live

Headless Browser

The agent runs a headless Chromium instance — it can navigate pages, capture screenshots, and automate web tasks.

live

Web Server (nginx)

This portfolio is self-hosted from a dedicated nginx container on the homelab — the site you're reading runs on my own hardware.

live

Public Access Layer

Evaluating a Cloudflare Tunnel for select public endpoints — with authentication in front before anything goes live.

in progress

Encrypted Offsite Backups

Proxmox Backup Server → nightly encrypted snapshots offsite. A proper 3-2-1 backup strategy.

planned

FIDO2 SSH Hardening

Hardware-key-backed SSH on critical nodes — physical touch + PIN required to log in.

planned

Google Integration

Read-only Gmail + Calendar access for the agent, so it can surface what matters without write access.

planned

02Weekly Log A running journal of what I ship each week.

● updated weeklynewest entries first

WEEK OF JULY 6, 2026

The agent gets hands, eyes — and this site goes live

Connected the AI agent to Home Assistant — it can now read device state and control the smart home through a scoped token.
Migrated the agent's reasoning to Claude Opus, with an automatic fallback model for reliability.
Gave the agent a headless browser — verified end-to-end with live navigation and automated screenshots.
Deployed this very site from a dedicated nginx container running on the homelab — self-hosted, serving live.
Stood up Cloudflare Tunnel infrastructure to build toward secure, authenticated remote access.

FOUNDATION · EARLY JULY 2026

Standing up the platform

Installed Proxmox VE 9 bare-metal on the home server and configured networking from scratch.
Deployed Home Assistant OS as a proper KVM virtual machine.
Built the private AI agent in its own LXC container, running as a persistent systemd service with a Telegram interface locked to me.
Configured Tailscale as an exit node + subnet router for private remote access — no ports forwarded to the internet.

03Hardware Two machines: one always-on server, one workstation that doubles as a lab.

NODE_01

// always-on

Mini Server

roleProxmox host
hypervisorProxmox VE 9
guestsHA OS · AI agent · more
networkingstatic IP · Tailscale
uptime24/7

NODE_02

// workstation + lab

Pop!_OS Rig

cpuRyzen 7 5800XT
gpuRTX 5070 · 12GB
memory32GB DDR4
storage2TB NVMe SSD
osPop!_OS (Linux)

04About Me 23 · Edison, NJ · turning operations experience into cloud engineering.

I'm Junior Batista, 23, based in Edison, NJ. I've spent 4+ years keeping high-volume infrastructure running in 24/7 Amazon operations — hardware break-fix, networking, and tier-1 IT support. I hold an A.S. in Computer Science and I'm currently pursuing a B.S. in Cloud & Network Engineering at WGU. BatistaCloud is where I put the theory into practice.

Process Assistant (L3)

Amazon · Edison, NJ

Primary point of contact for floor hardware and network issues — independently resolving ~80% of workstation, scanner, and printer tickets before IT escalation. Manage 50+ mission-critical handheld devices per shift in a 24/7 environment.

Jun 2024 – Present

Process Guide

Amazon · Edison, NJ

Handled workstation setup, laptop connectivity, and system diagnostics at shift start. Trained new associates on RF and finger scanners to reduce user-error tickets.

Jul 2022 – Jun 2024

05Toolbox & Education

// hands-on with

Skills

Proxmox / KVMLinux CLILXC Containers Virtual MachinesTCP/IPDNS / DHCP Tailscale / WireGuardCloudflare Tunnelsystemd Home AssistantSSHAI / LLM APIs Hardware Break-FixPythonJava

// credentials

Education

B.S. Cloud & Network Eng.WGU · exp. 2027
A.S. Computer ScienceMiddlesex · 2025

Coursework: Cloud Foundations, Network & Security, Scripting & Programming, Java & Python.