About

I grew up in Surabaya, East Java, and built my first real machine there: a humanoid robot head with seven degrees of freedom. I was a computer engineering student at PENS, and the head was my final project. I wrote the kinematic model, simulated it, and drove it from a desktop app over USB. My supervisor thought it was unusual enough to submit to an IEEE symposium, so at twenty-two I stood in front of an international room and explained a robot. I have been explaining machines to rooms ever since.

From Surabaya to the decacorns

After graduating I went to Jakarta and joined Tokopedia, where I learned what scale actually feels like: feeds, affiliate tracking, recommendation engines, gRPC services that could not be allowed to fall over. Then Jogja, first at Sorabel, where I kept the hot-path services alive and made a marketing broadcast run nine times faster with goroutines, and then at Ruangguru, where I led the backend for a chatbot’s first release and watched it become the channel that brought in the most new students that year.

Five years of making other engineers faster

In 2021 I moved to Bandung and spent five years at Evermos as lead platform engineer. My job was not to build features. It was to build the things that let everyone else build features: a search platform, a change-data-capture service, a WhatsApp messaging core, a CRM, role-based access control, finance tooling, and eventually the AI layer inside our chat product. Each one took a task that used to cost an engineer a week and made it cost an afternoon. That is still the work I am proudest of, and it is the habit I bring everywhere: find the thing everyone is rebuilding, and build it once, properly.

Sitting with founders

Somewhere in those years I started taking side projects where I was the only engineer in the room, talking to the founder directly. No product manager, no ticket queue, no deck between us. I heard the problem in their words, built the system, and then had to explain it back in words they could use to make a decision. That changed how I build. The best requirement I have ever received was a sentence a founder said out loud, and the most useful thing I can hand back is a plain explanation of what they now own. I still work this way, and I think more engineers should. It is also where the small-company end of my range comes from.

Agentic systems, on cloud and on-prem

I adopted large language models early. At Evermos I shipped a transactional chatbot on Azure OpenAI that creates orders through tool calling, and grounded answers with retrieval on Amazon Bedrock. From late 2024 I began working part-time with early-stage companies, taking them from experiments to agentic systems in production: a shopping assistant on Gemini through Google Vertex AI, a document analyzer for company policies that runs fully offline, and back-office routines that an agentic harness now runs on its own. In January 2026 I joined ZafarLabs, a Dubai-based team, as lead AI platform engineer, working remotely from Indonesia as a forward-deployed engineer for Dubai Government and enterprise clients. The pattern is the same at that scale: multi-agent platforms for citizen services, an air-gapped security operations platform, Arabic-first voice AI with a text-to-speech pipeline that can run in the cloud or on a client’s rack. The deployment target changes. The harness does not.

Why “leap higher”

It has been my GitHub motto for years. Four cities in Indonesia, one team in Dubai, one long diagonal. I do not think the point is to arrive anywhere in particular. The point is that the line keeps rising.