◦ Fine-tuning Vision-Language Models using HuggingFace for precise document form and table extraction, achieving a 3% improvement in accuracy.◦ Enhancing pond detection with satellite imagery and object detection, achieving 97% accuracy and saving 10 million IDR by reducing field visits.◦ Establishing a fraud identification service for shrimp ponds using image classification and Google Cloud Image Web Detection API, achieving 98% accuracy and saving 500 million IDR in costs.◦ Optimizing AI services by applying ONNX for inference, reducing production inference time from 6 seconds to 2 seconds.

◦ Developed several AI applications to prevent document spoofing for eKYC with 95% accuracy.◦ Designed a pipeline to enhance annotation speed for object detection applications using semi-supervised learning techniques, increasing data annotator productivity by approximately 35%.◦ Streamlined data labeling processes for Natural Language Processing use cases with Active Learning and HuggingFace, reducing data labeling duration process by 30% through advanced AI research focused on Active Learning with Clustering.◦ Collaborated with a team of 5 AI Engineers to design APIs, standardize documentation, and organize folder structures, simplifying project development and ensuring consistency.

◦ Collaborated with a team of 5 (2 Software Engineers, Project Manager, 2 AI Engineers) to develop AI products road map (Static Document Extraction and Product Shelf Recognition), successfully delivered to 13 clients and generated 495 million IDR in revenue.◦ Led a team of 2 AI Engineers in developing efficient AI solutions, achieving 98% accuracy for ProductRecognition and 97% for Static Document Extraction with 2-second inference time, reducing costs by 20%.◦ Established new AI models to extract six static documents (Tax Payer ID Card, Passport, Driving License, Vehicle Identity, Family Identity, and Children’s Card Identity) with 95% accuracy, acquiring six new clients and increasing revenue by 30%.◦ Coached 2 AI Interns in Deep Learning, Convolutional Neural Networks, Image Classification, Image Similarity, and Object Detection.

◦ Researched text detection, object detection, and one-shot learning to invent three AI applications (Product Shelf Recognition, Static Document Field Detection, Signature Matching) with 95% accuracy, reducing costs by 40%, using PyTorch, OpenCV, and NumPy.◦ Explored cutting-edge computer vision research, improving AI model performance and enhancing Static Document Field Detection accuracy by 3% with a 4-second CPU inference time.◦ Designed API documentation for 3 AI applications for production using Flask.

• Assisted in development of the front end of a web application for using HTML, CSS, and Javascript.• Applied Django framework to develop back end of a web application and integrated the API to front end.• Collaborated with 3 software engineers using version control systems such as Git to organize modifications and assign tasks.

• Supported lecturer in developing curriculum, teaching materials, exercises, and grading systems (quizzes, homework, andreports) for Introduction to Programming with Python course.• Taught around 30 students about Introduction to Python at the class.