Byron L.

Senior Data Scientist @BT Group

Working in the Next-Best-Action (NBA) squad Data Science Capability (DSC) squad to develop the multi-channels NBA platform and GCP ML DevOps solution from sketch, which can recommend the best action to customers with suitable channels and placements.\\\\n- Develop data pipelines, model training pipelines, and deployment pipelines (DevOps, Docker, CICD, Shell, Python); also documentation\\\\n- Create and train models, including reinforcement learning, graph convolutional network, and NLP\\\\n- Design, develop and deploy with teammates for several dashboards\\\\n- Test the feasibilities and create quick prototypes (Reinforcement Learning and Graph) for infrastructure on AWS/GCP\\\\n\\\\nAchievement\\\\n- Developed the end-to-end (MLOps) solutions (from data pipeline to endpoints) for the squads with Gitlab CICD, Docker, and AWS/GCP\\\\n- Built graph datasets, graph models (GCN with NLP), and graph reinforcement learning models for product recommendation\\\\n- Built several custom reinforcement learning models on StableBaseline3 libraries with Tensorboard\\\\n- Created a custom GYM environment on OpenAI GYM schema with the offline-validation approach\\\\n- Applied batch process for offline RL training, improved training speed by times, improved data pipelines\\\\n- Built several operational dashboards and Tensorboard-based training monitoring dashboard on Streamlit\\\\n \\\\n\\\\n \\\\n \\\\n\\\\n \\\\n