Ronaldyanto

Key Skills

Software

Label Studio

Top Subject Matter

Sports analytics (badminton stroke classification)

IoT deep learning (electricity meter digit recognition/OCR)

Top Data Types

3D Sensor

Image

Top Task Types

Action Recognition

Data Collection

Freelancer Overview

I have worked as a freelance machine learning engineer. Through this job, I got so much skills about data labeling and AI Training. I was worked on classification of empty and non-empty chair. During this project, I must labeling the image of empty and non-empty chair using label studio. I also have a project about multimodal deep learning model for badminton stroke classification and i was labeled the image based on the stroke type. I am also graduate from Mathematics with speciality on Machine Learning and Big Data. So, my thesis is also about training large language models.

Labeling Experience

Freelance Machine Learning Engineer — IoT deep learning projects

3D SensorData Collection

Developed and deployed deep learning projects for IoT devices, including OCR-based digit recognition for electricity meters that improved real-world data acquisition accuracy. Led an end-to-end ML workflow from data preprocessing through deployment for multiple production-bound deep learning deliverables. The role included managing ML pipeline execution to ensure timely delivery of trained models for edge/IoT use cases. • Developed OCR digit recognition for electricity meters using IoT-acquired data. • Preprocessed and prepared datasets for deep learning model training. • Built and managed end-to-end ML pipelines from data to deployment. • Led a team of three ML engineers and coordinated project delivery timelines.

2023 - 2025

Non-degree student exchange (TEEP NTUST) — multimodal deep learning model for stroke classification

3D SensorAction Recognition

Built a multimodal deep learning model using image and sensor data to classify badminton strokes. The work involved preparing and using labeled training data from both modalities to train an action recognition model. The project was evaluated in an academic setting where performance was graded and documented. • Used image and sensor streams for multimodal learning. • Trained an action classification model for badminton stroke categories. • Iterated on preprocessing and model development for better accuracy. • Achieved an A grade for related coursework and tutorial work.

2024 - 2024

Education

Work History