Projects
A selection of work across AI, software engineering, and automation.

01
Superfreak Studio
Custom 3D Printing & Digital Manufacturing Platform
Co-founded Superfreak Studio in 2025 with Bintang Timurlangit, starting as a 3D-printed furniture brand. In 2026, expanded into a full custom 3D printing service platform where users upload and manufacture their own 3D models directly through the website — a digital manufacturing workflow system for on-demand 3D printing services.
Role
Co-Founder & Full-Stack Engineer
Responsibilities
- Co-founded the studio and architected the full-stack manufacturing platform from the ground up
- Built the Next.js frontend and NestJS backend with TypeScript across the entire stack
- Developed Superslice, an internal slicing system for production workflows
- Integrated Midtrans payment gateway and RajaOngkir shipping calculation for automated commerce
- Designed cloud-based file upload and storage infrastructure using Cloudflare R2 for 3D assets
- Designed the web-based ordering flow for custom fabrication services inspired by industrial manufacturing platforms like PCBWay
Features
- Manufacturing File Pipeline — Cloud-based file handling with Cloudflare R2 for 3D model uploads, asset storage, and manufacturing file management
- Custom Slicer (Superslice) — Internal slicing system built to support production workflows and prepare uploaded models for manufacturing
- Commerce Integration — Midtrans payment gateway and RajaOngkir shipping calculation to automate ordering and fulfillment for Indonesian customers
- Production Workflow — Users upload custom 3D files, configure print settings, calculate manufacturing costs, place orders, pay, and track submissions
Challenges
- 01Designing a web-based ordering flow for custom fabrication services that translates industrial manufacturing UX into 3D printing
- 02Managing upload and storage pipelines for large user-submitted 3D assets
- 03Building internal slicing infrastructure to convert uploaded models into production-ready print configurations
- 04Maintaining a unified TypeScript architecture across frontend and backend for rapid iteration
Key Learnings
- 01Translating industrial manufacturing workflows (PCBWay-inspired) into a consumer-facing 3D printing platform
- 02Building a custom slicer (Superslice) for automated production preparation
- 03Integrating payment and logistics infrastructure for a full commerce-enabled manufacturing service
- 04Architecting full-stack TypeScript applications with NestJS and Next.js for maintainability
02
AI-Powered YouTube Shorts Automation
Autonomous AI Media Pipeline for End-to-End Video Generation & Publishing
A production-ready automation platform that transforms real-time news into fully rendered and published YouTube Shorts videos without human intervention. The system orchestrates news scraping, AI script generation, speech synthesis, subtitle synchronization, AI visual generation, FFMPEG rendering, and automated YouTube publishing through a configurable multi-channel automation architecture.
Role
Systems Engineer & Pipeline Architect
Responsibilities
- Architected end-to-end autonomous video generation pipeline from content ingestion to YouTube publishing
- Built configurable video profile system allowing dynamic customization of aspect ratio, resolution, subtitle positioning, AI visual behavior, scene composition rules, and rendering presets
- Migrated initial n8n prototype into a dedicated backend service architecture for scalability and maintainability
- Designed asynchronous queue workers to prevent concurrent FFMPEG jobs from saturating VPS resources
- Integrated ElevenLabs TTS, Whisper transcription, and Stable Diffusion into a unified multi-stage pipeline
- Implemented Google OAuth-based multi-channel YouTube publishing with dynamic profile routing
Features
- Content Acquisition — Scrapes real-time news/articles via Crawl4AI, extracts headlines and article content from trusted sources, triggers workflows via webhooks
- AI Content Generation — Converts news articles into short-form YouTube Shorts scripts, generates narration via ElevenLabs TTS, transcribes audio into word-level timestamps using Whisper
- Scene Orchestration — Splits scripts into scene-based segments, matches each segment to precise narration timestamps, generates AI visuals for every scene dynamically
- Video Rendering Pipeline — Uses FFMPEG to render scene clips based on exact segment durations, concatenates segments into final composition, burns dynamically positioned subtitles, synchronizes narration/subtitles/visuals automatically
- Publishing Infrastructure — Uploads videos automatically to YouTube, supports multiple YouTube channels through Google OAuth, routes automations dynamically based on selected video profiles
- Queue-Based Rendering — Designed asynchronous job queue workers to prevent multiple FFMPEG rendering jobs from running simultaneously on limited VPS infrastructure, preventing CPU/RAM exhaustion and pipeline slowdowns
- Fault-Tolerant Processing — Implemented stage-based processing and recovery mechanisms to prevent total pipeline failure when individual generation steps fail
Challenges
- 01Maintaining accurate alignment between AI narration, word-level subtitles, scene timing, and AI-generated visuals across the entire pipeline
- 02Optimizing long-running FFMPEG rendering jobs on constrained VPS infrastructure using queue-based concurrency control
- 03Integrating speech, language, transcription, rendering, and image generation systems into a coherent autonomous pipeline with recovery at each stage
Key Learnings
- 01Designing event-driven automation architectures that gracefully handle failures at any pipeline stage without losing progress
- 02Building resource-efficient rendering infrastructure with queue workers for production batch processing
- 03Engineering configurable, profile-driven systems that allow non-technical users to customize complex automation behavior
- 04Recognizing when to migrate from low-code prototypes (n8n) to dedicated backend services for scalability
03
Crypto Void
AI-Powered Crypto Media Automation
An automated crypto media pipeline that fetches latest crypto news, generates AI images via Flux, creates media-style posts, and auto-publishes to Instagram and Threads. Built with a NestJS backend service orchestrating n8n workflows for end-to-end content automation.
Role
AI Automation Engineer
Responsibilities
- Architected end-to-end content automation pipeline from news ingestion to social media publishing
- Integrated Flux image generation models for contextually relevant crypto visuals
- Built NestJS backend service to orchestrate n8n workflows and handle API integrations
- Implemented automatic scheduling and posting to Instagram and Threads platforms
Challenges
- 01Managing API rate limits across multiple social platforms while maintaining posting schedules
- 02Optimizing Flux generation costs while maintaining image quality and relevance to crypto news
- 03Handling diverse content formats (images, captions, hashtags) across Instagram and Threads
Key Learnings
- 01Designing fault-tolerant automation pipelines that gracefully handle API failures
- 02Orchestrating multiple AI services (image generation, news fetching) into a cohesive workflow
- 03Cost optimization strategies for AI API usage in production systems
04
UPT JPD Chatbot
WhatsApp AI Chatbot for Education Services
An intelligent WhatsApp chatbot built for UPT Jaminan Pendidikan Daerah (Dinas Pendidikan Pemuda dan Olahraga Kota Yogyakarta) to automate public inquiries. Powered by Deepseek as the LLM, orchestrated through n8n workflows, and connected via WAHA (WhatsApp HTTP API). Features include admin handoff with a 10-minute chat request window, Redis-based anti-spam, and smart filtering to ignore group chats and broadcasts.
Role
AI Automation Engineer
Responsibilities
- Architected end-to-end chatbot workflow using n8n for message routing and AI processing
- Integrated Deepseek LLM for context-aware responses to public education inquiries
- Implemented WAHA (WhatsApp HTTP API) for reliable WhatsApp connectivity
- Built admin handoff feature allowing users to request live chat with staff within a 10-minute window
- Designed Redis-based rate limiting to prevent spam and abuse
- Configured smart message filtering to bypass group chats and broadcast messages
Features
- Deepseek LLM for intelligent public inquiry responses
- Admin handoff with 10-minute live chat request window
- Redis-based anti-spam to prevent message flooding
- Ignores group messages and broadcast lists
Challenges
- 01Ensuring the admin handoff window reliably expires after 10 minutes without leaving sessions dangling
- 02Tuning Redis rate limiting to block spam while allowing legitimate rapid-fire questions
- 03Handling WAHA connection stability for 24/7 uptime
Key Learnings
- 01Designing production-grade WhatsApp chatbots with n8n workflow orchestration
- 02Implementing real-time admin handoff mechanisms within messaging platforms
- 03Building anti-spam systems with Redis for high-throughput messaging environments

05
EPICS In IEEE
An AIoT Based Smart Agricultural System for Pests Detection
EPICS (Engineering Projects in Community Service) is a global service-learning program by IEEE. We partnered with the Faculty of Agriculture at UGM to help salak farmers in Sleman, Yogyakarta encounter pest issues. We used IoT sensors to monitor temperature, humidity, light, and rainfall, and applied machine learning to predict weather and pest outbreaks. We also developed MySalak, a PWA web app that displays sensor data, prediction results, and weekly fly trap counts (FTD) across all farming areas.
Role
Full Stack Developer & IoT Engineer
Responsibilities
- Participated in designing and building an IoT device for environmental sensing used in this project.
- Conducted site surveys and installed 7 IoT sensors across 3 different plantation locations.
- Participated in developing the MySalak application and implemented the notification feature in the app.
Key Learnings
- 01Deploying IoT sensors in real agricultural environments with LoRa communication
- 02Building PWAs that serve farmers with limited connectivity in rural areas
- 03Cross-disciplinary collaboration between engineers and agricultural scientists
Project Stats

06
BirdTec
Agile Technica
A mobile application that is used to help chicken farmers to manage operationals like feed stock, live and dead chickens, and more
Role
Frontend Developer
Responsibilities
- Implemented core features for the plasma module, enabling effective registration and management of partner farms
- Worked closely with backend developers to ensure seamless API integration and data consistency across the plasma module
- Identified and fixed bugs to enhance app stability and user experience on both Android and iOS devices
- Actively contributed to Agile workflows through daily stand-ups, sprint planning, and regular retrospectives
Features
- Daily Unit Operation Management
- Plasma Management
- Feed Stock Management
Key Learnings
- 01Building offline-first mobile experiences for agricultural field workers
- 02Managing complex state in React Native with Zustand across multiple modules

07
TemanHR
Agile Technica
A cloud base mobile application to help companies simplify their HR and payroll processes.
Role
Frontend Developer
Responsibilities
- Implemented new UI components and screens based on design specifications using React Native and TypeScript
- Collaborated closely with backend developers and project managers to ensure seamless integration of features
- Debugged and resolved issues to ensure smooth functionality across both Android and iOS platforms
- Participated in Agile ceremonies including daily stand-ups, sprint planning, and retrospectives
Features
- Employee Attendance
- Overtime Management
- Timesheets Management
- Employee Leaderboards
- Automatic Payroll
Key Learnings
- 01Building enterprise-grade HR systems with complex business logic on mobile
- 02Designing responsive UI that handles large datasets (employee records, timesheets)