Senior SWE & AI/automation consultant. I build AI agents that do real work โ tool calling, MCP integrations, sandboxed execution, evals โ where model behavior, product UX, and backend reliability have to agree. Then I run them in production.
๐ Website ยท ๐ผ LinkedIn ยท ๐งต Threads ยท โ๏ธ Email
- โ๏ธ kilat.host โ Managed hosting: one-click deploys of AI agents, n8n, and PocketBase on Coolify. Flat billing from an IDR wallet topped up via QRIS.
- ๐๏ธ kilat.store โ Website builder and storefront surface of the Kilat platform.
- ๐ฌ aiclipper.video โ Serverless pipeline turning long-form video into vertical clips: transcription, AI segment selection, automated rendering.
- ๐ DIYVisa โ Privacy-first U.S. B1/B2 visa preparation workspace for Indonesian DIY travelers.
I fix things upstream when I can โ 59 of 70 authored PRs merged:
- rails/rails ยท heartcombo/devise ยท anthropics/claude-agent-sdk-python ยท can1357/oh-my-pi ยท sanger/sequencescape ยท benbalter/count-org-loc ยท benbalter/gman ยท phonology024/make-localllm-easier
A few things I maintain myself:
- ๐ฌ rails-mailer-sandbox โ Development email sandbox for Rails: persistent .eml storage and a Mailtrap-inspired web UI.
- ๐ค rails-ai-gateway โ Mountable Rails AI gateway with ActiveRecord, streaming, fallbacks, and an admin UI.
- ๐จ jamboard โ Collaborative infinite canvas: TypeScript, HTML5 Canvas, real-time multiplayer via Yjs, inline AI asset generation.
๐ก Signal โ ๐ Context โ ๐งญ Policy โ ๐ Tools/MCP โ ๐ฆ Sandbox โ ๐ฌ Feedback โ ๐ Eval
- ๐ Tools & MCP โ tool calling and MCP integrations with schemas that don't surprise
- ๐ฆ Sandbox โ code execution with guardrails, not vibes
- ๐ Context โ RAG, memory, retrieval, and traceable sources
- ๐งญ Policy โ route, reason, ask, answer, call a tool, or wait
- ๐ Observation โ traces, evals, review states, latency, and cost controls
The useful part is not an impressive internal chain of thought. It is an agent whose every action is inspectable, bounded, and evaluated.
๐ค AI systems Agentic workflows ยท Tool calling ยท MCP ยท Sandboxed execution ยท Evals ยท RAG ยท Vector databases ยท LLM orchestration
๐ป Product engineering TypeScript ยท React ยท Next.js ยท SvelteKit ยท Astro ยท Ruby on Rails ยท Node.js
โ๏ธ Data and infrastructure PostgreSQL ยท Redis ยท Docker ยท Terraform ยท AWS ยท Alibaba Cloud ยท Google Cloud
๐ Bekasi, Indonesia ๐ฎ๐ฉ ยท Working globally


