Challenge: Creating structured, complex surveys and quizzes manually takes significant time and effort.

Solution: Developed an intelligent platform that generates Google Forms instantly via multimodal input (Text/PDF/Images). Features interactive chat revisions and automated quiz generation, processing 100+ questions in under 3 minutes.

• Frontend: Next.js 16, React 19, Tailwind CSS 4
• Backend: Bun, Hono
• AI: Google Gemini 1.5 Flash-Lite
• Integration: Google Forms & Drive APIs, Tripay Payment Gateway
• Database: Firebase Firestore

Challenge: Superficial everyday digital interactions limiting deeper human connection.

Solution: Architected a full-stack platform serving 24,000+ users generating over $1,250+ monthly revenue. Features a custom prompt generator powered by the Gemini API and automated zero-downtime CI/CD deployment on GCP.

• Frontend: SvelteKit, Tailwind CSS
• Backend: Node.js, Express
• AI: Google Gemini API
• Cloud/DevOps: GCP Cloud Run, Cloud Build, Docker, CI/CD
• Database: PostgreSQL

Challenge: Diagnosing broken hardware devices and electronics from visual cues is difficult for average users.

Solution: Engineered an AI-powered visual assistant implementing a "Think-Search-Act" loop using Gemini 2.5 Flash-Lite and Google Search Grounding. Features server-side image annotation and automated step-by-step troubleshooting guide generation.

• Frontend: Next.js 16, React 19
• Runtime: Bun
• AI: Google Gemini 2.5 Flash-Lite & Search Grounding
• Image Processing: Sharp.js
• Cloud: Google Cloud Run

Challenge: Low English proficiency and the need for culturally relevant, accessible learning tools in Indonesia.

Solution: Led AI engineering for a multi-feature learning MVP. Utilized Microsoft Azure AI Services (Vision for object translation, OpenAI for dialogues) and designed an interactive, gamified learning experience with built-in Azure AI Content Safety.

• AI: Azure AI Vision, Azure OpenAI Service, Azure Content Safety
• Backend: Python (Flask)
• Cloud: Microsoft Azure Services
• Prototyping: Figma

Challenge: Inefficiencies and long processing times in the traditional zakat collection and distribution process.

Solution: Developed an AI application using Gemini for instant calculations and Whisper V3 for audio documentation. Created a national-level tracking dashboard. Enhanced zakat processing efficiency by 5x (from 10 mins to 2 mins).

• AI: Google Gemini API, OpenAI Whisper V3
• Backend: Python (Flask)
• Frontend: Streamlit (Dashboard)
• Automation: n8n

Challenge: Repetitive and time-consuming filling of end-of-semester evaluation forms for university courses.

Solution: Developed a Python desktop application using Selenium for browser automation and PyQt for the user interface, saving students significant manual entry time.

• Core: Python
• Automation: Selenium
• GUI: PyQt
• Packaging: PyInstaller

Challenge: Accessing and downloading open statistics data from Indonesia's BPS efficiently without complex portal navigation.

Solution: Created a lightweight, interactive Terminal User Interface (TUI) tool using Python to streamline exploration and bulk downloading of BPS datasets directly from the command line.

• Core: Python
• TUI Framework: Textual
• APIs: BPS Open Data API

Challenge: Fetching real-time ticket availability from train ticketing systems programmatically with bypass security layers.

Solution: Developed a high-performance scraping API featuring intelligent caching (TTL 15 min) and request routing to handle 30 req/min with 99.9% uptime.

• Core: Python (FastAPI)
• Scraping: Cloudscraper
• Containerization: Docker