A compressor predictive maintenance system developed for PT Astra Visteon Indonesia. It monitors plant telemetry across compressor, nitrogen, and dryer systems, pairing persistent anomaly alerts with breach-ETA forecasts and multi-week degradation trends.
02GenAI / RAG / FastAPI
InvenioAI
An advanced RAG system for PDF question answering, combining hybrid dense + sparse retrieval (Qdrant BM42), RAG Fusion, cross-encoder reranking, and structured Chain-of-Thought reasoning. Built with FastAPI, Streamlit, and Docker.
A CPU-native CLI and Python library for retrieving video moments from local videos using natural-language queries. MobileCLIP2 and Whisper-Tiny run through ONNX; SQLite FTS5 and sqlite-vec combine dialogue and visual matches.
An AST-scoped Python repair CLI that uses an LLM to fix a targeted function while preserving code outside its target AST node. Candidate patches are checked in isolated Git worktrees with test and lint gates before merging.
A histopathology classifier for lung carcinoma using a fine-tuned ConvNeXt-Base model (TensorFlow/Keras), with 93.67% test accuracy. Grad-CAM highlights tissue regions behind each prediction.
Data analytics internship at Tunas Group, on-site in Jakarta.
Analyzing automotive sales, inventory, and aftersales telemetry data across dealership networks.
Developing automated SQL queries, data pipelines, and executive dashboards for business metrics.
Sept 2024 – Present
Built a production RAG lab assistant on a fine-tuned Qwen3-8B, running on NVIDIA DGX.
Built a production-ready AI laboratory assistant that helps students troubleshoot laboratory systems and programming issues using a fine-tuned Qwen3-8B model and Retrieval-Augmented Generation (RAG).
Published peer-reviewed research applying deep learning (CoralNet, InceptionV3, MobileNetV2) to automate coral bleaching detection, achieving 89% accuracy to support marine biodiversity monitoring.
Deployed distributed AI workloads on NVIDIA DGX-1 and DGX A100 with infrastructure teams using Docker.
Jun 2026 – Sep 2026
Shipped the company's first AI predictive maintenance system, raising critical-event recall from 0% to 60%.
Built the company’s first AI-Based Compressor Downtime Monitoring and Predictive Maintenance System using React, FastAPI, Python, and SQL.
Analyzed 150K+ operational records and applied recall-focused optimization, boosting critical event detection recall from 0% to 60%.
Developed an end-to-end pipeline and web dashboard enabling real-time anomaly detection, automated alerts, and live operational monitoring for factory engineers.
Feb 2025 – Aug 2025
Taught GenAI, NLP, and LLM application design to international students and faculty.
Delivered hands-on training on Generative AI, NLP, and LLM applications to international students and academic faculty, bridging AI theory with practical implementation workflows.
Guided participants in architecting end-to-end GenAI applications, covering NLP basics, LLM integration, prompt engineering, and production-ready system design.
Jun 2025 – Jun 2026
Mentored 200+ participants through end-to-end ML projects, from data to deployment.
Mentored and assessed 200+ participants building end-to-end ML solutions, from data preprocessing through model deployment.
Developed practical AI/ML learning modules based on real-world engineering scenarios, emphasizing reproducible experimentation and production-oriented development.
Led technical reviews for participant projects: code review, debugging support, and system architecture recommendations.
Education
Sept 2021 – Aug 2025
Bachelor of Informatics
Universitas Gunadarma / GPA: 3.85 / 4.00
Studied computer science while actively building AI systems, publishing research, and teaching data science.
Feb 2024 – July 2024
Machine Learning Specialization
Bangkit Academy / Graduated with Distinction
Graduated with Distinction from Google-led ML program covering Deep Learning, Computer Vision, and NLP. Built CrashSnap as capstone, a YOLOv8-based vehicle damage detection system for automated visual inspection and damage assessment.
A few places where my work took shape.
From office floors to computing labs.
PT Tunas Ridean Tbk (Tunas Group)
PT Astra Visteon Indonesia
HPC Universitas Gunadarma
Technical skills
AI & Machine Learning
*Advanced RAG
*AI Agents
*Deep Learning
*Computer Vision
*NLP
*LLMs & GenAI
*Prompt Engineering
*PEFT & QLoRA
*Anomaly Detection
ML Frameworks & Libraries
*PyTorch
*TensorFlow
*scikit-learn
*LangChain
*Pydantic AI
*LlamaIndex
*Qdrant
*FAISS
*Hugging Face
Languages & Backend
*Python
*SQL
*TypeScript
*FastAPI
*Pandas
*NumPy
*Next.js
*React
*Git & GitHub
Cloud & MLOps
*Docker
*Microsoft Azure
*Google Cloud (GCP)
*MLOps Pipelines
*CI/CD
*NVIDIA DGX Systems
Spoken Languages
*Bahasa Indonesia (Native)
*English (Professional, TOEFL: 650)
Research
JITET · Vol. 13 No. 3
Model CoralNet, InceptionV3, dan MobileNetV2 untuk Klasifikasi Kondisi Terumbu KarangModel CoralNet, InceptionV3, dan MobileNetV2 untuk Klasifikasi Kondisi Terumbu Karang
