About the Role
We are seeking an Applied Data Scientist to join our Algorithm team — a hands-on builder at the intersection of LLM systems, agentic AI, and crypto-native intelligence.
You will design, build, and improve AI agents that understand markets, reason over on-chain data, and take actions at speed for hundreds of millions of users — across reasoning models that plan before they act, multi-agent systems that orchestrate complex crypto workflows, and test-time scaling that pushes inference-time intelligence to its limits.
You build for real environments: designing agent architectures, improving performance across quality and latency, developing evaluation methods for open-ended tasks, and turning research ideas into production.
Responsibilities
- Design and implement production-grade LLM pipelines powering Binance AI Products and next-generation agentic trading features — including multi-step reasoning agents, tool-selection frameworks, and autonomous workflow execution across spot, perpetual, and on-chain markets
- Continuously improve agent capabilities in understanding, reasoning, tool selection, and action execution — optimizing simultaneously for intelligence, latency, and reliability under high-frequency trading constraints
- Build and maintain evaluation frameworks for reasoning model outputs in crypto contexts — covering market analysis accuracy, agent decision quality, hallucination detection, and adversarial robustness against prompt injection in financial workflows
- Apply test-time scaling techniques — chain-of-thought, self-consistency, process reward models — to push agent reasoning quality in ambiguous, fast-moving market conditions
- Architect AI system components with rigorous attention to inference latency, throughput, and cost efficiency — leveraging serving frameworks such as vLLM and TensorRT-LLM — in a zero-downtime, 24/7 trading environment
- Integrate on-chain data sources, wallet intelligence, and crypto market signals into LLM-powered analytical pipelines — building the data layer that makes Binance's agents genuinely crypto-native
- Partner with research scientists to translate experimental findings into production-grade agentic solutions with clear performance benchmarks
Qualifications
- Bachelor's or Master's degree in Computer Science, Electrical Engineering, Statistics, Mathematics, or related technical field
- 0–5 years of industry or research experience in applied ML or AI engineering
- Strong Python programming skills ; Equally important: demonstrated comfort with vibe coding — using AI-assisted development tools fluidly as core part of your workflow
- Demonstrated hands-on experience with LLMs — prompt engineering, post-training, or end-to-end LLM application development
- Familiarity with multi-agent system design — task decomposition, tool use via MCP, memory management, parallel agent execution, and inter-agent communication
- Strong analytical thinking and problem decomposition; comfortable operating under ambiguity in fast-moving environments