Airline pilot
ζ©εΈ« Γ θ»ι«ε·₯η¨εΈ«,ηΎε¨η¨δΈζ― AI agent θ¦ιζ code δΊ€δ»εΊε»γ
Β πΉπΌ
I've shipped software since 2010. Over the past year my workflow changed shape: I design the system, set the guardrails, and let autonomous coding agents handle implementation, pull requests, reviews and infra β end to end. My recent commit history is increasingly agent-authored, human-reviewed.
The point isn't fewer engineers β it's a higher altitude. I spend my time on architecture and judgment; the mechanical work is delegated and verified.
flowchart LR
J["π§ββοΈ Jonathan<br/><sub>architecture Β· guardrails Β· review</sub>"]
subgraph FLEET["π€ Autonomous coding fleet"]
direction TB
A["Feature<br/>agent"]
B["Infra / IaC<br/>agent"]
C["Data & backtest<br/>agent"]
D["Quality / CI<br/>agent"]
end
J -- intent & guardrails --> FLEET
A --> O["π¦ PRs Β· deploys<br/>pipelines Β· green CI"]
B --> O
C --> O
D --> O
O -- review & merge --> J
- 1min-cloudflare-gateway β high-performance Cloudflare Workers API gateway
- apple-store-scrape β cross-region Apple price comparison (Python + Workers)
- pdf-processor β browser-side PDF toolchain (PyMuPDF β WASM)
- finlab-txd β quantitative trading research on Taiwan index futures
Self-hosted, auto-updated daily. Aggregate counts across all my repos β private repo names are never exposed. Commits classified by author & co-author trailers.