I'm building AI automation, SaaS systems, and high control software for real operators.

Founder of To Stupid Too Quit LLC in Phoenix, AZ.

Most of what I'm building is private inside my org right now. GitHub is where I'll publish the cleaned up parts: architecture notes, case studies, templates, and useful open source artifacts.

home-service-hero is a pre alpha operations command center for residential trade businesses, including HVAC, plumbing, electrical, and field service teams.

Current state:

• Placeholder page
• Next.js 15
• TypeScript strict
• Vitest scaffold
• ADR driven roadmap

Planned direction:

• Auth
• Multi tenancy
• Scheduling workflows
• Quoting workflows
• Customer follow up automation
• Housecall Pro connectors
• QuickBooks connectors
• MCP and AI assisted operations

Real product work is gated on accepting ADRs 0002 and 0003. I'm not trying to fake traction off a placeholder page. I want the foundation right first.

polymarket-oracle-bot is a Python 3.11+ async trading bot for directional BTC positions on Polymarket prediction markets.

It prices fair value using dual Chainlink BTC/USD oracles with The Graph subgraph cross verification.

Core rules:

• Half Kelly position sizing
• 2% max position cap
• Minimum 9% edge before entry
• Minimum $500k market liquidity
• Formal state machine for execution and monitoring

State machine:

IDLE -> SCANNING -> ANALYZING -> EXECUTING -> MONITORING

Failure branches:

SAFE_MODE
SHUTDOWN

Risk controls:

• Emergency USDC.e sweep to cold multisig
• Drawdown shutdown above 10%
• Gas spike protection above 40 gwei
• Oracle disagreement handling
• Fernet encrypted state
• HMAC signed JSON logs
• Prometheus metrics on :8000 and :8001
• Dockerized non root runtime
• Monte Carlo backtester targeting less than 0.001% ruin probability

This repo is private because trading systems shouldn't be public until the safety story is boring.

dgm-hunter is a self evolving, hexagonal architecture hyperagent platform for autonomous bug bounty research.

The system has a Meta Agent that rewrites the Task Agent's instructions using Darwin Gödel Machine style feedback from hunt metrics. The Task Agent runs the loop across recon, vulnerability detection, and validation.

Attack engines include:

• SSTI
• IDOR and race conditions
• CRLF
• HPP
• OOB testing
• Subdomain takeover checks
• 10 other focused engines

Safety and validation:

• scope_policy.yaml gate before findings move forward
• Sanitizing MCP tool gateway
• ProveRAG validation against NVD and CWE
• Ed25519 signed provenance
• Local Ollama runtime with cloud fallback
• Redis and Qdrant through Docker Compose
• 1166 tests
• 96% coverage
• Zero ruff findings
• Zero mypy findings

The goal isn't a hacker movie agent. It's controlled autonomy with scope enforcement, evidence, provenance, and repeatable validation.

I like:

• Clear system boundaries
• Explicit state machines
• Boring failure modes
• Audit trails
• Useful logs
• Tests that catch real regressions
• Risk controls before scale
• AI that's boxed in by policy, validation, and permissions

I don't like:

• Demo ware
• God mode agents
• Mystery automation
• Fake enterprise polish
• Sloppy systems hiding behind big words
• Pretending a placeholder page is a product

• AI assisted operations for service businesses
• Secure RAG and agent systems
• SaaS architecture for small teams
• Workflow automation
• Trading system risk controls
• Bug bounty automation with scope enforcement
• IBM watsonx and hybrid cloud

Most active repos are private today. I'm working toward public artifacts that are actually useful:

• Home Service Hero architecture notes
• ADRs from real builds
• Secure agent workflow patterns
• Polymarket bot risk control case study
• DGM Hunter architecture case study
• SaaS starter patterns for service businesses

Website: tostupidtooquit.com
GitHub: @02ez
Org: To Stupid Too Quit
LinkedIn: linkedin.com/in/02ez