Binance Accelerator Program

About the role

The Binance Accelerator Program offers university students and recent graduates a 3-6 month internship to gain practical experience within the global digital asset industry. Participants join the Binance AI Pro team to contribute to the development of agentic systems and AI-native experiences.

Key facts

What you'll do

• Develop and refine Agent Runtime components such as intent routing, RAG, query rewriting, and multi-step workflows.
• Create and optimize tool-calling systems, including discovery, execution, and integration with internal services and market data.
• Build evaluation datasets and testing frameworks to assess agent reliability, answer quality, and routing accuracy.
• Perform failure analysis and optimize workflows, prompts, and guardrails to enhance agent performance.
• Implement system monitoring, logging, and tracing to ensure observability and reliability.
• Troubleshoot asynchronous pipelines and manage integration testing using mocked LLM responses.
• Collaborate with algorithm, data science, and backend engineers to move AI features from experimentation to production.

Requirements

• Proficiency in Python and strong software engineering fundamentals.
• Solid grasp of data structures, algorithms, and system architecture trade-offs.
• Practical experience with LLM applications or agent systems via projects, research, or prior internships.
• Knowledge of RAG, function calling, agent workflows, and LLM evaluation techniques.
• Ability to navigate production codebases, CI/CD, APIs, and asynchronous service environments.
• Analytical debugging skills with a focus on identifying root causes of system failures.

Nice to have

• Experience with frameworks like LangGraph, LangChain, AgentScope, or LlamaIndex.
• Knowledge of MCP, tool ecosystems, or agent runtime environments.
• Background in building LLM testing infrastructure or benchmarks.
• Familiarity with vector search, embeddings, or LLM observability tools.
• Experience with Kubernetes, Docker, Kafka, or Redis.
• Exposure to AI safety, prompt injection defense, or model guardrails.
• Understanding of cost optimization and model inference latency.

Skills & tools

• Python
• LLM Agent Frameworks
• RAG
• Tool Calling
• CI/CD
• Asynchronous Services
• System Observability
• AI Evaluation

Practical notes

This role provides a competitive salary and company benefits. Work-from-home arrangements are available based on specific business team requirements. All applicants must review the Candidate Privacy Notice before submission. Terms of engagement are subject to local laws and contract agreements.