Machine Learning Engineer with hands-on production experience at Telkom Indonesia and a public portfolio of 20+ end-to-end ML projects with live demos. Built and deployed NLU, NER, and TTS systems powering a production chatbot platform; fine-tuned LLMs with QLoRA/LoRA/DoRA; built RAG and multi- agent LLM systems; and shipped full MLOps pipelines (Airflow, dbt, MLflow, Docker, CI/CD, drift monitoring). Informatics graduate, Summa Cumlaude (GPA 3.92/4.00), Telkom University.
– Engineered the NLU intent-classification pipeline of the Mona production chatbot through 3 model generations: TF-IDF + SVM, fastembed + ONNX (bge-small-en-v1.5), and a fine-tuned IBM Granite multilingual embedding model; augmented training data with Gemini Flash and Gemma, curating ~2,000 high-quality samples from 5,000 raw. – Developed an end-to-end Named Entity Recognition (NER) pipeline in 2 generations (rule-based to custom SpaCy model trained on domain data), integrated across all 4 service layers: frontend builder, backend, engine, and NLU. – Built a text-to-speech (TTS) voice engine b
– Implemented anomaly detection on the ELK Stack (Elasticsearch, Logstash, Kibana) to flag unusual threat occurrences on specific source IPs. – Built a multi-device login detection mechanism and a Telegram alerting pipeline for detected anomalies.
Kampus Merdeka Study Independent program – Completed 500+ hours of structured ML training and 35 certifications (DeepLearning.AI, Google, Dicoding). – Capstone: Harvest Scan, a plant-disease detection mobile app using MobileNetV2 transfer learning with 95% validation accuracy.