Binance Accelerator Program - AI Backend Engineer

Binance is a leading global blockchain ecosystem behind the world’s largest cryptocurrency exchange by trading volume and registered users. Binance is trusted by more than 320 million people in 100+ countries for its industry-leading security, transparency, trading engine speed, protections for investors, and unmatched portfolio of digital asset products and offerings from trading and finance to education, research, social good, payments, institutional services, and Web3 features. Binance is devoted to building an inclusive crypto ecosystem to increase the freedom of money and financial access for people around the world with crypto as the fundamental means.

About Binance Accelerator Program Binance Accelerator Program (BAP) is a 3-6 month internship program designed for Early Career talent to have firsthand experience in the rapidly expanding digital assets space. You will be given the opportunity to develop your skills at Binance and understand what it’s like to work at the world's leading blockchain ecosystem. As part of your internship in the BAP, there will also be opportunities for networking and development, which will expand your professional network and build transferable skills to propel you forward in your career. Learn about the BAP Program HERE.

Who may apply Current university students and recent graduates. *Terms of employment / engagement shall be subject to contract and local applicable laws

What you'll do: Build and improve pieces of the agent runtime: intent routing, query rewrite, RAG/FAQ answering, tool/skill orchestration (L0–L6 loop). Extend the Tool Server — add and optimize tools (web search, market data, report APIs), improve tool-call efficiency and output quality. Strengthen evaluation: write eval datasets and harnesses, measure intent-routing accuracy and answer quality, catch regressions before release Improve observability: traces, metrics, logging (Opik/Pinpoint), production dashboards and alerts.Fix bugs across the pipeline (routing, skill loading, guardrail false-triggers) and write integration tests with mock-LLM cassette replay.

Must-have: Solid Python and clean coding habits; comfortable in a real service codebase with CI, code review, and typing. Understanding of LLM application basics — prompting, RAG, tool/function calling, or agent frameworks (from coursework, projects, or internships). Able to reason about correctness and write tests; debug across async/services.Available for a sustained internship and eager to iterate quickly on feedback.

Nice-to-have: Experience with an agent framework (AgentScope / LangChain / LlamaIndex or similar). Familiarity with vector search / embeddings, evaluation frameworks (Opik, ragas), or LLM observability.Exposure to Kafka, Redis/Lindorm, S3, config systems (Apollo), or Docker/K8s. Interest in evaluation, guardrails/safety (prompt-attack defense), or latency/cost optimization.What you'll get Ship real features into a production AI agent used across the app. Mentorship from senior BE/DS/algo engineers; end-to-end ownership of scoped tasks. Hands-on exposure to the full modern LLM-app stack: RAG, tool servers, evals, LLMOps.