.png&w=256&q=75)
Sandhika Hamzah@rinesro
1
Events attended
1
Submissions made
Indonesia
1 year of experience
About me
Sandhika Hamzah is a fresh graduate of the Informatics program at Universitas Gunadarma, where he finished with a GPA of 3.76. He is based in Depok, West Java, and is open to placements outside the city. His main interest is turning raw data into something people can actually act on. For his thesis he built EnergiCerdas AI, a web platform that analyses household energy consumption and returns practical recommendations for reducing it. Before that he worked in a team to deliver an inventory and asset management system for Kelurahan Gedong, a real government client with real operational requirements. He has also built Lumière Gallery, a solo web application using CodeIgniter 4 and MySQL with authentication, a masonry image gallery, and threaded comments. Alongside project work, he has completed several independent data case studies covering student dropout prediction, bike sharing usage patterns, and retail sales analysis, using Python, SQL, and Looker Studio. He holds an IDCamp Data Scientist certification at Expert Level with an Outstanding predicate, has passed the BNSP Data Engineering competency assessment, and has completed data analysis training through Dicoding. His plan is to build a career step by step, starting in data analytics and visualization and moving toward data science and AI engineering. He enjoys work that lets him dig into a problem on his own, ship something that runs, and then improve it.
🤓 Latest Submissions
Onboarding Assistant
The problem When a developer joins a new project, the first days usually go to reading code instead of contributing. They keep asking the same three questions: Where do I start? How do I run it? What can I safely change? The answers usually live in a senior teammate's head, because onboarding docs are slow to write by hand and go stale quickly, so they are often skipped. The solution Onboarding Assistant turns any unfamiliar codebase into a ready-to-read onboarding guide with one IBM Bob skill run. It produces four documents: an architecture overview (purpose, tech stack, key components, folder tree, data flow), Mermaid class and sequence diagrams built from the real models and endpoints, a step-by-step setup guide with troubleshooting, and five starter tasks that each point to the exact files a newcomer should touch. A small script then renders them into one static website that can be opened locally or deployed anywhere. Target users and how they use it The tool is for development teams that onboard new members, open source maintainers who want to lower the barrier for first contributors, and developers who inherit an unfamiliar or legacy repository. Usage takes three steps: open the repository in Bob IDE, switch to the "Onboarding Guide" mode, and run /onboarding-assistant. The result is a site the whole team can share. What makes it creative and unique Bob is the engine of the workflow, not a side helper. The solution is built entirely from Bob's own building blocks: A custom mode that can read the whole repository but can only write documentation, so source code is never changed by accident. A reusable skill that encodes the full analysis workflow, so any team gets the same structure on any repo. Four parallel subagents, one per document, each working in an isolated context. This keeps the main conversation small and reduces cost. A deliberate split between AI and code: AI does the understanding, while a zero-dependency Node script does the rendering. Rebuilding the site after an edit costs zero Bobcoins and takes under a second. How it effectively and efficiently addresses the problem We ran the same skill on two unrelated stacks to show it is not tied to one project: IBM's Galaxium Travels (Python FastAPI and React) and JobGap (Next.js and TypeScript). Both produced complete onboarding sites. Testing the second repository also exposed a permission limit in the custom mode, which we fixed so it now works at any folder depth. Measured from the Bob task log, one normal run took about 6 minutes and about 2.3 Bobcoins, compared with the hours it usually takes to write onboarding docs by hand. Because the output is plain markdown plus a static page, teams can review it, edit it, and regenerate it at no extra AI cost. New developers get an architecture map, a working setup path, and a safe first task on day one, and senior developers spend less time repeating the same explanations.
27 Sep 2026
