Hi, I'm blockedby

Senior Backend & Applied AI Engineer focused on agentic developer tooling, production backends, and practical automation.

I build systems where LLMs and coding agents are not just chat interfaces, but bounded, inspectable engineering tools.

What I work on

• Coding-agent workflows and developer productivity tooling
• Codex / Pi / Claude-style agent integrations
• Deterministic web evidence + LLM summarization
• Backend systems in Go, TypeScript, PostgreSQL, Redis
• Production debugging, observability, CI/CD, and deployment automation
• FinTech / DeFi / Web3 infrastructure

Selected work

Agentic engineering lab

Public index of my AI tooling, agent workflows, OSS contributions, and sanitized production case studies.

Repo: https://github.com/blockedby/agentic-engineering-lab

pi-codex-tools

Pi extension exposing Codex CLI as bounded tools for deterministic web search/fetch, delegated coding tasks, patch validation, structured JSON outputs, debug logs, and timeouts.

Repo: https://github.com/blockedby/pi-codex

Obscura browser runtime contribution

Rust browser-runtime contribution fixing full-load navigation hangs on JS-heavy pages by adding bounded script/event-loop execution and regression tests.

PR: h4ckf0r0day/obscura#195

Hermes Agent gateway work

Fork/work around Hermes Agent messaging gateway behavior, including Telegram-based agent access workflows.

Repo: https://github.com/blockedby/hermes-agent

Go OpenRouter SDK work

Go SDK work around OpenRouter / LLM API features such as streaming, reasoning, tool calling, structured outputs, prompt caching, multimodal inputs, and usage fields.

Repo: https://github.com/blockedby/go-openrouter

How I work

I like small, sharp PRs with clear acceptance criteria, verification, and evidence.

I also maintain Pi Agent Setup, a public Pi/Codex agent-stack bootstrap for bounded engineering workflows: owner/implementer/auditor routing, reusable skills, browser/Codex integrations, explicit secrets boundaries, and smoke-test verification.

Typical loop:

1. define the problem;
2. split it into bounded slices;
3. use agents for exploration, implementation, review, and tests;
4. keep deterministic evidence separate from model output;
5. verify with tests/builds/static checks;
6. write down what changed and why.

Background

Previously worked across high-load backend systems, DeFi/Web3 infrastructure, smart contracts, fintech workflows, CI/CD, and team leadership.

Main stack: Go, TypeScript/Node.js, PostgreSQL, Redis, Docker, Kubernetes, Linux, Solidity.

Open to Applied AI Engineer, AI Tooling Engineer, Forward-Deployed Software Engineer, Backend AI Platform, and Founding Engineer roles.