Paul Bertens Email & Phone Number

Netherlands

Seoul, South Korea

Performed fundamental research on LLMs, GNNs, and created EXAONE Discovery, a platform for interacting with ML models through LLMs to assist in the discovery of new materials.► Full-stack development: frontend + backend + databases + ML models. Interface using ReactJs+NextJs, backend APIs using FastAPI for accessing machine learning models and DB interactions (postgress). ► LLM fine-tuning: prompt engineering for data generation + implementation for full-parameter fine-tuning… Show more Performed fundamental research on LLMs, GNNs, and created EXAONE Discovery, a platform for interacting with ML models through LLMs to assist in the discovery of new materials.► Full-stack development: frontend + backend + databases + ML models. Interface using ReactJs+NextJs, backend APIs using FastAPI for accessing machine learning models and DB interactions (postgress). ► LLM fine-tuning: prompt engineering for data generation + implementation for full-parameter fine-tuning LLMs. Split task into conversation agent optimized for chat interactions, and tools agent optimized for interacting with external APIs.► Fundamental Research GNNs: Development of graph-based diffusion model for forward reaction prediction to predict potential reaction impurities, and provide interpretable predictions.► Fundamental Research LLMs: Development of multi-modal graph+text models, to integrate chemistry knowledge into graph-based approaches, better grounding predictions and improving overall prediction accuracy. Show less

Seoul, Korea

Fundamental research on Brain-Inspired AI, using knowledge of Reinforcement Learning (RL), Evolutionary Algorithms, and Computational Neuroscience. ► Published in NeurIPS: "Emergence of Hierarchical Layers in a Single Sheet of Self-Organizing Spiking Neurons." Advances in Neural Information Processing Systems (NeurIPS), 2022► Published on the cover of Nature Machine Intelligence: "Network of evolvable neural units can learn synaptic learning rules and spiking… Show more Fundamental research on Brain-Inspired AI, using knowledge of Reinforcement Learning (RL), Evolutionary Algorithms, and Computational Neuroscience. ► Published in NeurIPS: "Emergence of Hierarchical Layers in a Single Sheet of Self-Organizing Spiking Neurons." Advances in Neural Information Processing Systems (NeurIPS), 2022► Published on the cover of Nature Machine Intelligence: "Network of evolvable neural units can learn synaptic learning rules and spiking dynamics." Nature Machine Intelligence, 2020 Show less

Tokyo, Japan

Developing YOKOZUNA data, a scalable AI platform applying Machine learning to game data in order to predict individual player behavior. ► Management: Leading development of big data architecture, moving to cassandra and distributed database design, expanding kubernetes cluster for scalable web application. ► Big Data Engineering: Applying spark and other techniques for efficient data processing, parallelizing algorithms and machine learning models across multiple cores and… Show more Developing YOKOZUNA data, a scalable AI platform applying Machine learning to game data in order to predict individual player behavior. ► Management: Leading development of big data architecture, moving to cassandra and distributed database design, expanding kubernetes cluster for scalable web application. ► Big Data Engineering: Applying spark and other techniques for efficient data processing, parallelizing algorithms and machine learning models across multiple cores and machines. ► Software Engineering: Developed overall system architecture and database design, push notification systems, push automation, sending items to users in-game and implemented front-end + back-end of new features. ► Research: Researched and published papers in top conferences on new models and techniques. ► Sales/ marketing: Attended international conference and meetings, acquiring new customers. Show less

Within 23 Wards, Tokyo, Japan

Developing a scalable AI platform applying Machine learning to game data, in order to predict individual player behavior. ► Research: Participated and won churn prediction competition CIG2017, published in international conferences. ► Machine Learning: Developed and implemented Deep Learning and ensemble models for predicting player behavior.► Front-end engineering: implemented web application front-end using html, js and css for visualizing machine learning model… Show more Developing a scalable AI platform applying Machine learning to game data, in order to predict individual player behavior. ► Research: Participated and won churn prediction competition CIG2017, published in international conferences. ► Machine Learning: Developed and implemented Deep Learning and ensemble models for predicting player behavior.► Front-end engineering: implemented web application front-end using html, js and css for visualizing machine learning model results, and allowing interaction with the platform for taking actions to retain players. ► Big Data Engineering: Applied big data processing and storage techniques for scaling to large datasets. Implemented and deployed on distributed systems on the cloud using kubernetes and docker images. Show less

Deep Learning applied to music information retrieval (MIR)

Doctor Of Philosophy - Phd, Brain And Cognitive Engineering

Master Of Science (Msc), Artificial Intelligence, Cum Laude

Exchange Program, Big Data

Bachelor Of Science (Bsc), Computer Science