Research Data Scientist

This is a fully remote position, open to applicants in Singapore.

📋 Description

• Assist in the design and implementation of experiments related to reasoning model training, post-training alignment, test-time scaling, or comprehensive model evaluation, particularly focusing on applications within financial and crypto-native environments.

• Analyze and summarize the latest academic research from NeurIPS, ICML, ICLR, and ACL, keeping track of advancements in reasoning, alignment, and agentic AI to refine and direct ongoing research efforts.

• Develop model variants and training methodologies, including those based on RLVR, as well as evaluation protocols utilizing PyTorch and the Hugging Face ecosystem.

• Systematically monitor and document experiments using Weights & Biases or a similar tool, ensuring adherence to reproducibility standards.

• Investigate the convergence of LLM reasoning and crypto-native data, such as on-chain indicators, market microstructure, and multi-modal market intelligence, while pinpointing unique research opportunities that align with Binance's strategic position.

• Work in collaboration with applied engineering teams to comprehend how research insights translate into production limitations in a continuous, 24/7 trading environment.

⛳️ Requirements

• Currently enrolled in a Master's or PhD program in Machine Learning, Computer Science, Mathematics, or a related discipline is highly preferred.

• Anticipated graduation in 2026, 2027, or 2028.

• Proficient in Python programming and experienced with PyTorch; familiarity with C++ or Rust is advantageous. Equally significant is a demonstrated ease with vibe coding—effectively using AI-assisted development tools as an integral part of your research and experimentation process.

• Strong grasp of transformer architectures, large language model pretraining, and the progression toward reasoning models.

🏝️ Benefits

• Competitive salary and comprehensive company benefits.

• Work-from-home option (specific arrangements may vary based on the operational needs of the business team).