Pioneer Talent Program - Research Data Scientist

Binance is a leading global blockchain ecosystem behind the world’s largest cryptocurrency exchange by trading volume and registered users. We are trusted by over 230 million people in 100+ countries for our industry-leading security, user fund transparency, trading engine speed, deep liquidity, and an unmatched portfolio of digital-asset products. Binance offerings range from trading and finance to education, research, payments, institutional services, Web3 features, and more. We leverage the power of digital assets and blockchain to build an inclusive financial ecosystem to advance the freedom of money and improve financial access for people around the world.

About the Pioneer Talent Program  Are you an early-career professional ready to kickstart your career in the dynamic world of blockchain? Binance’s Pioneer Talent Program is designed to nurture ambitious individuals like you. The program offers full-time roles where participants will gain hands-on expertise via a tailored training curriculum and personalized 1:1 mentorship from industry leaders. Dive behind the scenes of the world’s leading blockchain ecosystem while building the skills to thrive in Web3. Launch your career with purpose—join us in shaping the future of digital innovation.

Who May Apply Candidates with up to 5 years of relevant work experience

About the Role We are seeking a Research Data Scientist who combines rigorous research thinking with a drive to see findings land in production. You will conduct original research in generative AI and LLMs applied to crypto-native problems — spanning reasoning under uncertainty, agent reliability in adversarial financial environments, and multi-modal market intelligence. This is a research-first role at the intersection of three critical frontiers: post-training alignment and reasoning for financial agents, efficient inference and test-time scaling, and multi-agent coordination in open, adversarial environments. The goal is real-world impact — not just papers, but agents that trade, analyze, and act for millions of users. You formulate research questions, design rigorous experiments, build evaluation frameworks, and collaborate with engineers to bring research into production.

Responsibilities Conduct original research in generative AI and large language models applied to crypto and financial domains — with focus areas including reasoning under market uncertainty, alignment of trading agents, multi-agent coordination, and test-time scaling for time-critical decisions Design and execute rigorous experiments — formulating clear hypotheses, implementing model and training variants including RLVR-based reasoning approaches, running systematic ablations, and drawing statistically sound conclusions Develop novel post-training methodologies and evaluation frameworks for LLMs operating in crypto contexts — covering chain-of-thought quality for market analysis, agent decision consistency, and robustness against adversarial prompt injection Research test-time scaling techniques — process reward models, self-consistency, Monte Carlo Tree Search-based planning — and apply them to improve agent reasoning quality in ambiguous, fast-moving market conditions Critically track developments across the research community at NeurIPS, ICML, ICLR, ACL, and crypto-adjacent venues — identifying high-leverage opportunities to apply state-of-the-art techniques to Binance's unique challenges Qualifications Master's or PhD in Machine Learning, Computer Science, Mathematics, Statistics, or related field strongly preferred; exceptional Bachelor's candidates with demonstrable research output will be considered 0–5 years of research or industry experience in ML/AI; strong academic lab or research internship experience equally valued Deep understanding of transformer architectures, large language model pretraining dynamics, and post-training methodology — including the shift from RLHF toward RLVR-based reasoning model training Proficiency in Python and PyTorch, Critically, you work in an AI-native way — using vibe coding practices with tools like Claude Code, Cursor, or Copilot Workspace as your primary development workflow Rigorous mathematical foundations: linear algebra, probability theory, information theory, stochastic processes, and numerical optimization Bilingual English/Mandarin is required to be able to coordinate with overseas partners and stakeholders