I build useful software from messy inputs: papers, public data, rough ideas, and unfinished workflows.

Repositories | ScholarScout | pddikti-scraper | Indonesia

I like building things that start with a vague problem and end with a tool someone can actually use.

That usually means work around applied AI, research workflows, automation, internal tools, and product ideas that are practical enough to survive contact with real users.

Lately, that has looked like:

• building ScholarScout, a project that turns academic papers into directions for research, product, and feature ideas
• building pddikti-scraper, a utility for collecting and cleaning public PDDikti student search results from NIM patterns
• using repos like worldline and eureka as places to test product systems, interface ideas, and implementations while they are still taking shape

• software that is useful before it is flashy
• systems that stay clear even when the underlying problem is messy
• technical work that does not collapse into overengineering
• ideas that can move from prototype to something real

• experiments that turn information into something actionable
• tools built around research, automation, and decision support
• small products and utilities with a strong practical angle
• work in progress that is trying to become finished work

• thoughtful collaboration on AI tools, research systems, and internal software
• product ideas with a real use case behind them
• conversations with people who like building practical things