Data Scientist Intern

kumparan

Jul 2025 - Nov 2025

• Improved the Subchannel Prediction Model by adopting mBERT, raising the F1-Score from 90% (IndoBERT) to 92%.
• Executed a large-scale backfill workflow, predicting and updating 2M+ article channel & subchannel labels from 2017–2025.
• Collaborated as one of two Data Scientists in a 6 person cross functional team to develop an automated Topic Suggestion feature, benchmarked YAKE (72%) vs TF-IDF (75%) and selected TF-IDF. Built an asynchronous model inference pipeline using FastAPI and structured BigQuery integration for backend deployment. Awarded Best Final Project.

Data Scientist Intern

PT Telkom Indonesia Tbk

Oct 2024 - Mar 2024

• Designed and maintained interactive dashboards using Looker Studio to track key metrics.
• Processed MyIndiHome iOS App Review and Twitter data using n-gram analysis and LLM APIs for automated sentiment scoring, topic prediction, and summaries.
• Developed a Prompt-Based AI system using LLMs (ChatGPT, Gemini, LLaMA) to improve complaint data exploration efficiency by up to 50%.
• Generated over 32,000 Q&A pairs for internal chatbot training.

Data Analyst Intern

Pusat Pengelolaan Komplek GBK

Jul 2024 - Aug 2024

• Analyzed 2023 e-booking data and visualized insights in a Streamlit dashboard to provide actionable recommendations for optimizing venue rentals.
• Analyzed sentiment from 65,000+ crawled tweets (April–June 2024), visualizing insights on a Streamlit dashboard to support public complaint analysis and management. (3-member team project)

Bootcamp Mentee (DS & AI)

Dibimbing

Jan 2026 - Jul 2026

• Developed deep learning models for image classification and implemented YOLO for object detection, utilizing Label Studio to annotate and curate custom datasets.
• Developed RAG systems powered by OpenAI models and Langchain, implementing ChromaDB and PostgreSQL (pgvector) to optimize document embedding, indexing, and context aware response generation.
• Built and evaluated supervised and unsupervised models (classification, clustering, time series forecasting) with focus on business-driven insights such as churn analysis.

Machine Learning Cohort

Bangkit Academy by Google

Feb 2024 - Jul 2024

• Graduated with average passing grade of 95.
• Developed a ML model to detect decent/unworthy clothes using TensorFlow MobileNetV2, achieving 94% accuracy.
• Served as project manager in a 7-person team to build the "WearShare" app, created detailed reports and designed the business workflow.