I'mArifian.

Automation tool built with **Python** and **Playwright** to streamline mentor grading workflows on the Infinite Learning LMS platform. The system automates authentication, navigation, assessment grading, comment submission, and validation processes, significantly reducing repetitive manual tasks and improving operational efficiency.

Major upgrade of the Arifian.AI personal chatbot. Rebuilt from the ground up with a modern RAG (Retrieval-Augmented Generation) architecture. Now powered by **GPT-OSS 20B via Groq** for fast, high-quality responses, with **multilingual SentenceTransformer embeddings** (paraphrase-multilingual-MiniLM-L12-v2) supporting 50+ languages including Indonesian. Features include a dedicated **admin dashboard** for knowledge base management, system prompt customization, file uploads, RAG configuration, and user management. Backend built with **FastAPI** and **MongoDB Atlas** vector search, hosted on **Hugging Face Spaces** with Docker.

Arifian.ai is a personalized **my-own-AI-persona chatbot** designed to answer general or public questions about myself. Made from **fine-tuned Google T5-small** with my own datasets and USE (Universal Sentence Encoder) for Feature Extraction. Hosted as a Spaces instance to serve API backend functionality in HuggingFace Spaces.

This project explores the use of LSTM (Long Short-Term Memory) and LSTM with an Attention Layer to predict daily rainfall rates in Batam City. The primary goal is to assess the performance difference between the two models when the Attention Layer is implemented. The study leverages the BMKG DataOnline climate dataset and aims to improve prediction accuracy, especially in the context of dynamic and extreme weather changes.

Developing a generative AI project using AMD PC AI technologies for the international **AMD Pervasive AI Developer Contest** with **Hackster**, in collaboration with a team of IBM Academy: Advance AI mentors from Infinite Learning. Open-Source pre-trained models used in this projects are : **OpenAI Whisper** as **Automatic Speech Recognition (ASR)** Pre-trained Model, **MarianMT** as **Machine Translation (MT)** Pre-trained Model, **tacotron2-DDC** as **Text to Speech (TTS)** Pre-trained Model, and **HiFi-GAN** as **vocoder** Model. Device used in this development is: Minisforum Venus UM790 Pro with AMD Ryzen™ 9 with Ryzen 9 7940HS as a hardware sponsorship.