Skip to content
View MADPANDA3D's full-sized avatar

Block or report MADPANDA3D

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
MADPANDA3D/README.md
█   █  ███  ████  ████   ███  █   █ ████   ███  ████  ████
██ ██ █   █ █   █ █   █ █   █ ██  █ █   █ █   █     █ █   █
█ █ █ █████ █   █ ████  █████ █ █ █ █   █ █████  ███  █   █
█   █ █   █ █   █ █     █   █ █  ██ █   █ █   █     █ █   █
█   █ █   █ ████  █     █   █ █   █ ████  █   █ ████  ████

        SYSTEMS ONLINE // HUMAN AT THE CONTROLS

Leo Lara

Founder · Systems Architect · U.S. Marine Corps Veteran · Cybersecurity Student
I turn scattered tools into owned, observable systems that do real work.

Founder of MADPANDA3D U.S. Marine Corps Veteran Cybersecurity Student


leo@madpanda3d:~$ mission
Build the system I could not find.

leo@madpanda3d:~$ operating_model
AI expands execution bandwidth.
Judgment, ownership, and verification stay human.

leo@madpanda3d:~$ status
Founder mode: active.
Systems: online.
Work: compounding.

whoami

I did not arrive at software through a straight line.

I spent four years as an Aircraft Rescue Firefighter in the U.S. Marine Corps, then moved into telecom emergency-response and break/fix work supporting T-Mobile. That path eventually led to nearly two years in an AWS engineering role.

Every chapter taught me the same lesson:

A system is only impressive if it keeps working when conditions stop being ideal.

In early 2024, a 3D printer turned a long-running idea into MADPANDA3D. What began with 3D design and early commissions expanded into web development, Linux, cybersecurity, network defense, automation, private-cloud infrastructure, and agentic systems.

Today I run MADPANDA3D as a veteran-owned technology agency and operate as a solo systems architect with an AI agent stack around me. I do not use AI to avoid thinking. I use it to increase execution bandwidth across research, design, engineering, QA, deployment, documentation, and operations—while keeping a human accountable at every important boundary.

I am also enrolled in National University’s Cybersecurity program. The deeper I go into security, infrastructure, business, and the human side of systems, the more convinced I become that the real work is integration: connecting the parts, making state visible, and leaving behind something another person can actually operate.

system_architecture

MADPANDA3D is not a collection of disconnected experiments. The long-term mission is one owned operating stack.

flowchart TB
    L["LEO / operator"] --> M["MADPANDA3D / one operating stack"]

    M --> P["Products + marketplaces"]
    M --> A["Agentic systems + MCP"]
    M --> I["Private cloud + observability"]
    M --> B["CRM + automation + client delivery"]

    P --> K["Filesystem-first knowledge + memory"]
    A --> K
    I --> K
    B --> K

    classDef core fill:#0B1120,stroke:#E11D48,color:#FFFFFF,stroke-width:2px;
    classDef lane fill:#111827,stroke:#22D3EE,color:#FFFFFF;
    classDef knowledge fill:#111827,stroke:#F59E0B,color:#FFFFFF;

    class L,M core;
    class P,A,I,B lane;
    class K knowledge;
Loading

active_systems

Some of my most ambitious work is private by design. These descriptions show the problems I am solving without exposing the infrastructure behind them.

System Current mission State
J.A.R.V.I.S. + KYBER A local-first agent stack combining voice, orchestration, memory, tools, and operator-visible sessions with a Linux-native desktop workbench. ACTIVE / PRIVATE
Charter A captain-first South Florida marketplace bringing discovery, verified onboarding, bookings, payments, QR tickets, and secure transfers into one marine platform. LAUNCH-STAGE
Home Lab + SOC/NOC A self-hosted command and observability plane with live telemetry, incident workflows, audit trails, approval gates, and provider-readiness views. OPERATIONAL
MAD MCP Portal An authenticated connector control plane that lets agents discover and route governed tools, services, skills, playbooks, and support workflows. DEPLOYED
MADPANDA Knowledge OS A filesystem-first system that turns business, academic, technical, and AI-conversation evidence into durable notes, source maps, journals, and scoped retrieval. OPERATIONAL

selected_public_work

Project What it demonstrates
QDRANT-MCP A managed vector-memory layer for agents with dry-run mutation previews, bulk maintenance, validation, quality controls, and authenticated hosted access.
GHL-MCP 269+ documented tools across 19+ categories, with multi-tenant CRM routing, safe-mode writes, health checks, and self-tests.
DISCORD-MCP A Java/Python automation service for moderated Discord operations, async jobs, scheduled audits, and confirm-gated actions.
FFMPEG-MCP Preset-only asynchronous media processing with validation, queues, signed downloads, templates, brand kits, and social exports.
MAD Card Exchange A proprietary, visual-first collecting platform whose public showcase includes demos, screenshots, and 101 public releases.

The public lab also includes HOSTINGER-MCP, CF-MCP, GOOGLE-MCP, ARCH-KITTY-CONFIG, and HYPRLAND-THEMES.

engineering_doctrine

  • Own the stack. Critical workflows should be understandable, observable, replaceable, and under operator control.
  • AI needs boundaries. Specialized agents get scoped tools, explicit routes, and human approval at high-risk edges.
  • Reliability beats theater. Dry runs, tests, health checks, logs, readback, rollback paths, and honest failure states are product features.
  • Build from the root cause. One fix at the shared boundary is better than patches scattered across every caller.
  • Knowledge should compound. Decisions and evidence belong in durable systems—not trapped inside one chat, dashboard, or person’s memory.
  • Ship the smallest complete unit. Working, verified, and maintainable beats impressive-looking complexity.
Toolbox // expand

TypeScript Python Rust Java Bash

React Node.js FastAPI Tauri Three.js

Docker Linux Proxmox Cloudflare Grafana

PostgreSQL Redis Qdrant Supabase Obsidian

open_channel

I work best on problems that cross boundaries:

  • product + automation
  • AI agents + governed tools
  • software + infrastructure
  • CRM + web + operations
  • private cloud + observability
  • scattered knowledge + one source of truth

I especially care about helping small businesses and veteran entrepreneurs gain ownership of their technology instead of inheriting another pile of subscriptions.

If your current stack is five tools, three spreadsheets, two automations, and no trusted source of truth, that is exactly my kind of problem.

Start a conversation Explore MADPANDA3D Connect on LinkedIn

BUILD THE SYSTEM // OBSERVE THE SYSTEM // OWN THE SYSTEM
MADNESS, DISCIPLINE & PANDA POWER.

Pinned Loading

  1. QDRANT-MCP QDRANT-MCP Public

    This is a project that allows you to work with a MCP that fully allows you to manage your own vector database. This is designed to work with N8N agents and give them the ability to extend what is p…

    Python 3

  2. GHL-MCP GHL-MCP Public

    This is a comprehensive GHL MCP with multi tenant capability and full readiness with an extensive funnel suite and forms / automations boost

    TypeScript 4 2

  3. MAD-CARD-EXCHANGE-PUBLIC MAD-CARD-EXCHANGE-PUBLIC Public

    1