The duty of the man who investigates the writings of scientists, if learning the truth is his goal, is to make himself an enemy of all that he reads, and … attack it from every side.
— Ibn al-Haytham
If, in some cataclysm, all of scientific knowledge were to be destroyed, and only one sentence passed on to the next generations of creatures, what statement would contain the most information in the fewest words? I believe it is the atomic hypothesis (or atomic fact, or whatever you wish to call it) that all things are made of atoms...
— Richard Feynman
Greetings, I'm Resha!
I am a research engineer with extensive experience in artificial intelligence and robotics, specializing in computer vision and machine learning. My work includes publishing peer-reviewed research on face recognition, object detection, and biomedical image segmentation, as well as contributing to the development of efficient AI models for both cloud-based systems and edge devices. In addition to my core expertise, I maintain a strong interest in natural language processing and life sciences.
I am open to opportunities in AI research, applied machine learning, and computer vision engineering.
🎓 Education
-
B.Eng., Electrical Engineering, Universitas Gadjah Mada, Yogyakarta (Indonesia), 2016
-
Thesis: NightOwl: Robotic Platform for Wheeled Service Robot.
-
Summary: NightOwl is a robotic platform designed exclusively for a wheeled service robot. It was developed for the receptionist service robot of the Electrical and Information Engineering Department of Universitas Gadjah Mada as part of the capstone project for the bachelor's degree graduation requirement. The robot will welcome guests and guide them to the desired location.
-
-
Thesis: NightOwl: Robotic Platform for Wheeled Service Robot.
-
M.Eng., Electrical Engineering, Universitas Gadjah Mada, Yogyakarta (Indonesia), 2022
-
Thesis: Convolution-Involution Combination Encoder-Decoder Network for Efficient Gastrointestinal Polyp Segmentation.
-
Summary: ConvInCEDNet, a novel lightweight gastrointestinal polyp segmentation model that can run efficiently on a low-end CPU-only device, particularly the Raspberry Pi 4, was proposed. It combines the strength of convolution and involution in a U-shaped encoder-decoder fashion. It was trained and tested on the CVC-ClinicDB, CVC-ColonDB, Endotect, ETIS-LaribPolypDB, KvasirCapsule-SEG, and Kvasir-SEG datasets. It shows competitive performance compared to previous models.
-
-
Thesis: Convolution-Involution Combination Encoder-Decoder Network for Efficient Gastrointestinal Polyp Segmentation.
-
Ph.D., System Information Sciences, Tohoku University, Sendai (Japan), 2025
💼 Work Experience
-
Research Assistant, UGM AI CENTER, Yogyakarta (Indonesia), Jun 2019 - Dec 2020
-
Developed back-end of automated machine learning pipeline for data scientists to increase productivity. The back end is written in Python with API built upon FastAPI.
-
Devised service robot platform. The platform is based on ROS. It manages streams of information that are processed and used for motion planning, SLAM, and human-robot interaction (HRI) of the robot.
-
Helped in writing and publishing scientific papers.
-
-
AI Engineer (Intern), Neurabot, Yogyakarta (Indonesia), Dec 2020 - Mar 2021
-
Developed a web-based auto-tagging system for medical image labeling. It will segment the salient objects that are present on the scene. The auto-tagging model is based on DEXTR.
-
Devised micro-scanner firmware for medical imaging. The firmware is built upon PlatformIO. The micro-scanner is intended to perform whole slide image acquisition.
-
-
Junior Machine Learning Engineer, Techbros, Remote (Germany), May 2021 - Mar 2022
-
Designed and maintained elangai, an AI inference engine platform for edge devices. The framework offers CLI commands that create and execute a project template, convert an optimized deep learning model into a consumable file, and inspect the model. The framework is written in Python.
-
Created and built AI projects using elangai. The projects mostly are computer vision problems, e.g., plant disease classification, face mask detection, face recognition, and license plate recognition. These projects are deployed on edge devices, especially NVIDIA Jetson.
-
-
Research Assistant, UGM AI CENTER, Yogyakarta (Indonesia), Jul 2022 - Dec 2024
-
Conducting research on knowledge distillation for medical image segmentation in robotics colonoscopy.
-
Proposed novel architectures for medical image segmentation. The architecture is composed of Transformer- and CNN-based components. Its performance is evaluated on three different datasets concerning medical image segmentation tasks, i.e., medical instrument segmentation, lesion segmentation, and anatomical structure segmentation.
-
Helped to develop a deep learning model for monocular depth estimation. The model is a modification based on U-Net. Its performance is validated on KITTI and NYU Depth V2.
-
Involved in research on a novel lightweight model for segmentation of unstructured off-road images. The model is trained and evaluated on ORFD, RUGD, and RELLIS-3D.
-
Carried out a project on geospatial image segmentation with UNetFormer. Images are taken from a low-altitude aerial vehicle.
-
Implementing ensemble machine learning models on large-scale data in the insurance industry. The models are maintained and deployed with orchestrated microservices. The models are trained at a scheduled interval. The prediction results are visualized and interpreted through the informative graphical UI.
-
Helped in writing and publishing scientific papers.
-
Providing reviews, recommendations, and discussions to the employees of an oil and gas company about data science platforms and their use in the industry.
-
This website is powered by Jekyll + Hyde. The source code is on github.