Yogyakarta, Indonesia Open to opportunities

Welcome, I’m Dana.

AI Engineer · Data Scientist

Third-year Information Technology student at Universitas Gadjah Mada, passionate about computer vision, NLP, and on-device AI.

Pradana Yahya
Yogyakarta, ID

01 - About

Curious by default, shipping by habit.

I like building AI systems that move from experiments into real use: fine-tuning small models that run on a Raspberry Pi, building RAG pipelines that need reliable answers, and training vision systems for robot soccer.

Recent work includes an offline BISINDO translator for the Tuli community (KawanIsyarat), a multimodal retail-analytics platform that won DATATHON 2025 (TRACKO), and a lightweight Indonesian text-to-speech model (NusaVoice).

Outside the terminal, I hike, play football, and lead Gadjah Mada Robotic Team - FUKURO, which qualified for RoboCup 2026 in Incheon, South Korea.

02 - Selected work

Featured projects.

03 - Experience

Where I’ve been.

  1. Dec 2025 - Jan 2026

    Yogyakarta, Indonesia

    Machine Learning Intern

    BMKG - Badan Meteorologi, Klimatologi, dan Geofisika

    Developed a PM2.5 air quality forecasting system with Random Forest and deployed it as a full-stack web app at bmkg-hirup.pradanayahya.com. Also built rainfall forecasting models from climatology station data and internal tools for BMKG workflows.

    scikit-learnFastAPIReactDocker
  2. Dec 2025 - Present

    Yogyakarta, Indonesia

    Research Assistant - Smart Lab & Smart Space

    DTETI, Universitas Gadjah Mada

    Research on AI and IoT systems for Smart Lab and Smart Space. Developed NusaVoice, a lightweight Indonesian text-to-speech system based on fine-tuned Piper/VITS models for CPU-only inference.

    PiperVITSFastAPIReact
  3. Oct 2025 - Dec 2025

    Yogyakarta, Indonesia

    AI Engineer Intern

    DataIns - PT Global Data Inspirasi

    Built Smart Space (Jarvis), a voice assistant for laboratory automation at DTETI UGM. The system runs ASR and TTS locally on a Raspberry Pi with Whisper and a fine-tuned Piper voice, then connects LLM command understanding to N8N workflows for lab device control.

    WhisperPiperLLMN8N

04 - Toolbox

Technical Skills.

Programming Languages

PythonTypeScript / JavaScriptC / C++JavaSQL

Machine Learning & AI

PyTorchTensorFlow / TFLiteHugging FacePEFT / LoRAscikit-learnPandas

NLP & LLM

IndoBERTRoBERTaBERTopicRAGAgentsWhisper · Piper

Computer Vision

OpenCVTimeSformerDINOv3ROS2TFLite on-device

Frontend

ReactNext.jsSvelteKitTailwind CSS

Backend & Data

FastAPIFlaskNode.jsPostgreSQLMongoDB

DevOps & Cloud

DockerGitAzureHome AssistantN8N

Hardware

Raspberry PiEdge devices