Edbert HalimAI Engineer

Proposed and implemented a hybrid topic modeling pipeline using BERTopic and Small Language Models (SLMs) to cluster documents and generate interpretable topic metadata.

Designed and developed a scalable RAG (Retrieval-Augmented Generation) pipeline for Djarum’s internal and external chatbots, leveraging a Docker-orchestrated backend stack (Postgresql, Redis, RabbitMQ, Elasticsearch) to enhance intelligent knowledge retrieval and conversational AI capabilities.

Developed key user-centric features such as a role-based access control system for exclusive content and an automated reminder system using email (SMTP) and Discord webhooks, scheduled via cron jobs, significantly improving user engagement and timely notifications.

Engineered a robust pipeline using scheduled workers and API integrations to continuously fetch user progress data from external platforms in real time. Optimized scraping strategy by cutting unnecessary API calls and reducing infrastructure usage by over 80%, while integrating with a custom analytics model to dynamically visualize learning progress.

Integrated Vue.js frontend with a scalable Django backend, focusing heavily on API design, performance optimization, and user-focused innovation.

Developed SQL queries to extract, clean, and analyze customer data, contributing to CRM performance reports and business insights.

Created and updated charts to visualize key CRM metrics, such as customer retention and engagement, providing actionable insights.

Preparing high school students for the Indonesian National Olympiad in Informatics. Focusing on advanced algorithms, problem-solving techniques, and competition strategies.