I'm Benedictus Kent Rachmat

• Scholarship recipient from SaclAI and Chalearn. ⭐
• Research training in Large Language Models and challenge organization with Prof. Isabelle Guyon. ⭐
• Research training in Graph Neural Networks under Prof. Francois Landes.
• Mentor to M1 students.
• Knowledge in Deep Learning, Machine Learning, Computer Vision, Natural Language Processing, Generative Models, Signal Processing, Reinforcement Learning, and more.
• Learned Chinese and Spanish
• Active in fencing

• Developed a range of applications, including multi-games, chatboxes, social media platforms, marketplaces, etc. ⭐
• Experienced in leading and collaborating on diverse projects, demonstrating strong team-building and project management skills.
• Knowledge in Java, C, networking, databases, Git, full-stack development, AMPL, Haskell, game development, cryptography, and more.
• Engaged in student organization activities.

Working on Large language model for code generation

Mentoring the M1 students

Prepararing for the next challenge class

Worked on the Auto-Survey Challenge in collaboration with Google which was presented at the AutoML Conference 2023 and JDSE conference 2023. My primary focus was on paper generation using LLMs and automated prompt engineering to create systematic overview papers in various domains. Leveraging state-of-the-art LLMs.

Used GNNs to predict molecular movements within glass structures

Designed and automated test cases with the help of Cypress and Zephyr.

Created/validated task tickets and bug tickets

We built the website's front-end and improved the database structure. We also made a system to help organize events and changed how things looked and worked on the site.

I helped to design and create their internal website using CodeIgniter along with JavaScript, CSS, HTML, JQuery, Bootstrap and the database. In total I have completed 3 different kind of project.

In the paper, we introduce an automatic system designed to streamline the review process of survey papers by treating it as a classification challenge. Our system aims to evaluate the relevance of scientific papers based on a specific prompt, using a unique dataset to train machine learning models for identifying the most pertinent papers. Preliminary results demonstrate that our BERT-based classifier notably outperforms traditional machine learning approaches.

In an ambitious collaboration with Google, I had the opportunity to work on the Auto-Survey Challenge under the supervision of Isabelle Guyon. This project was presented at the prestigious AutoML Conference and JDSE Conference in 2023. The core objective of the Auto-Survey Challenge was to innovate the process of generating and reviewing systematic overview papers in various academic and research domains. Leveraging the advanced capabilities of Large Language Models (LLMs). This initiative represented a significant stride in automating aspects of academic research, combining the latest in AI technology with scholarly needs.

I have contributed to Google's Gson community. Gson is a renowned Java library for converting Java Objects into JSON and back

Check out my github repo for my school projects ;)