ClarkKitchen

Teacher by day. Builder by night.

Shipping systems that work when reality shows up.

SYSTEM ACTIVE
DEPLOYED
BUILDING:
AI/ML
01/10
KYZLO: THE ORIGIN ─────

From classroom chaos to distributed systems. From late-night Discord calls to Kyzlo Labs. No pitch decks. No investors. Just building what needed to exist.

Built by Kyzlo

From the classroom to the command line. Systems that work when reality shows up.

Teaching is Systems Management

Chapter 01

Load balancing thirty kids' attention. Routing information through different channels for different learners. Handling the exceptions—the kid who needs something different—without crashing the whole system.

The patterns transfer directly to distributed systems, multi-agent orchestration, and infrastructure.

Pattern Recognition: Systems

11:47 PM, Tuesday

Chapter 02

Two friends in a Discord call, screens glowing in dark rooms miles apart. "Why does nothing work the way we need it to?"

Every tool we tried was built for someone else. Too bloated. Too generic. We'd duct-tape solutions together and waste hours fighting software instead of building what we actually wanted.

Discord
Late Nights
Frustration

What if we just... made it ourselves?

Chapter 03

No pitch decks. No investor meetings. No five-year roadmap.

Just a shared screen, a fresh repo, and two people tired of waiting for someone else to solve problems we understood better than anyone.

$ git init kyzlo-labs
$ git commit -m "First commit: Build for builders"

Built for Builders

Chapter 04

We build for people who think in systems. The ones who want control, not hand-holding. The ones who'd rather have something powerful and a little rough around the edges than something polished and limiting.

If you've ever wished a tool existed that just got how you think—welcome. We're still building. Still in that Discord channel at hours we probably shouldn't be.

Still Building
CODENAME:RAG BRAIN
complete
CLASSIFICATION: ARTIFICIAL INTELLIGENCE
PRIMARY INTEL:

Persistent memory layer for AI agents. Four coordinating agents handle quality gating, retrieval, retraining, and maintenance. The system learns what makes memories useful through feedback loops and self-retraining.

ASSETS DEPLOYED:
PythonPostgreSQLpgvectorXGBoostOllamaDocker
FIELD RESULTS:
  • Self-retraining quality model using agent feedback
  • Four-agent coordination pattern for autonomous maintenance
  • MCP integration for seamless agent memory
  • Wasserstein distance-based quality scoring
MISSION CONTEXT:
Technical: Vector similarity search with supervised learning...
Social: Understanding what humans actually need from...
SWIPE TO NAVIGATE
EQUIPMENT WALL ──────

// tools forged for real constraints

PERSISTENT MEMORY CORE

Agents forget. This remembers.

STORING...

EDGE DETECTION ARRAY

Finds what markets misprice.

SCANNING...

DATA SANITIZATION UNIT

FERPA-compliant. Deterministic.

SANITIZING...

MODEL FUSION KIT

Zero to fine-tuned in 48 hours.

MERGING...

VOCABULARY MODULE

Speaks the language of professionals.

Aa
PARSING...

POLITICAL SIMULATION ENGINE

The social layer, made playable.

SIMULATING...

// hover to inspect

$ SYSTEM OVERVIEW

Real-time capability monitoring

LANGUAGES

LanguageProjectsFilesLinesUsage
PythonPRIMARY
714218.4k
TypeScriptPRIMARY
58912.1k
SQL
423890
Shell
3151.1k
Rust
1122.3k
Solidity
181.2k
HCL
16340
>

AI/ML FRAMEWORKS

LangChain
80%
TensorFlow
65%
PyTorch
60%
OpenAI API
90%
Hugging Face
55%
>

WEB TECHNOLOGIES

React/Next.js
95%
Node.js
85%
Tailwind CSS
90%
PostgreSQL
70%
GraphQL
60%
>

INFRASTRUCTURE & DEVOPS

Docker
85%
Kubernetes
65%
AWS
75%
GitHub Actions
80%
Terraform
50%
>

ACTIVE PROJECTS

AI Agent Platform

Multi-agent orchestration system

2024-01-15

Web3 Dashboard

Real-time blockchain analytics

2024-01-10

ML Pipeline

Automated data processing

2024-01-08

API Gateway

Microservices orchestrator

2024-01-05
ABOUT: THE PROFILE ─────
0+
Projects Shipped
0+
HuggingFace Downloads
0
Production Deployments
CIS @ Ball State
Education

>WHAT_SETS_ME_APART

Systems Thinking

Computer Information Systems degree from Ball State with a consulting capstone. Classroom management is systems management. Load balancing attention, routing information, handling exceptions, maintaining state. The patterns transfer directly to distributed systems, multi-agent orchestration, and infrastructure.

Social Layer Awareness

Every system is a social system. The technical solution that ignores the human layer fails. RAG Brain learns what memories are useful to humans. Rome.Life makes you manage senator relationships. The sports model accounts for public betting psychology.

Learning Velocity

Zero Python to LoRA fine-tuning pipeline over a few months of consistent after-work long nights teaching myself. The classroom taught that understanding beats shortcuts. Learn the underlying systems, not just the APIs.

Ship Everything

Every project is deployed and running. Rome.Life for students. PSJanitor for school data. Trading bots on mainnet. If it's not shipped, it's not real.

TECH_STACK<

Languages

PythonTypeScriptSolidityRust

AI/ML

PyTorchTransformersLangChainOllama

Web

Next.jsReactTailwindPrisma

Infrastructure

DockerTerraformPostgreSQL
Building under the Kyzlo Labs brand
──────LET'S CONNECT──────

Let's Connect

Open to collaborations, interesting projects, and conversations about building systems that respect both the technical and human layers.

"Every system is a social system."
[Start a Conversation]