Data Scientist - Large Language Model

Responsibilities

• Advance Reasoning & Planning: Enhance reasoning and planning abilities in foundation models across all stages—data acquisition, model evaluation, supervised fine-tuning (SFT), reward modeling, and reinforcement learning (RL).
• Data Synthesis: Create and curate large-scale, high-quality datasets through techniques such as rewriting, augmentation, and generation, supporting improvements in pretraining, SFT, and RL phases.
• Complex Problem Solving: Tackle challenging tasks using system 2 thinking and advanced decoding methods such as Monte Carlo Tree Search (MCTS) and A*.
• Evaluation Methodologies: Design and implement robust evaluation frameworks to measure model capabilities and performance throughout development.
• Tool Use & Agent Development: Enable models to utilize external tools, interact with APIs, and run code interpreters. Develop both single and multi-agent systems to address sophisticated problems.

Requirements

• Bachelor’s, Master’s, or PhD in Computer Science, Applied Math, Statistics, or related fields.
• Solid research experience in reinforcement learning (RL), large language models (LLMs), and familiarity with large-scale model training practices.
• Strong foundations in data structures and algorithms, with proficiency in Python or C++/Java.
• Demonstrated impact through notable projects or publications in RL, natural language processing (NLP), or deep learning.
• Excellent analytical and problem-solving skills, especially in the context of large-scale model development and deployment.
• Strong communication and collaboration abilities, with a passion for innovation and advancing AI technologies with the team.
• Bilingual English/Mandarin is required to be able to coordinate with overseas partners and stakeholders.