You will be responsible for production-grade AI data processing and knowledge services for Binance’s equities business. You will transform financial information—announcements, news, earnings reports, earnings call transcripts and audio/video, research reports, and more—into data, knowledge, and evidence that trading products and Binance AI can directly consume. Knowledge engineering, knowledge bases, and retrieval-augmented generation (RAG) are the core scenarios. You will also build reusable processing frameworks and own model and rule integration, task orchestration, servitization, quality control, cost management, and production stability—not just document parsing, vectorization, or model API calls.
Own production-grade AI data processing pipelines for financial content (announcements, news, financial reports, earnings call materials, research reports, etc.), covering text, table, layout, and audio/video processing, as well as chunking, deduplication, clustering, standardization, versioning, and result validation.
Build layered financial knowledge bases that manage source documents, structured facts, entities and events, full-text and vector indices, and relationship data; maintain metadata including source, time, security, market, language, version, and authorization.
Engineer and operate RAG services in production—building query processing, permission and time filtering, multi-route retrieval, reranking integration, context assembly, evidence citation, and result return pipelines that ensure traceability to original sources.
Design extensible AI data processing, knowledge engineering, and indexing frameworks with unified interfaces to adapt to new sources, formats, languages, models, rules, and algorithms; support incremental updates, index rebuilds, historical backfill, deletion, and authorization expiry.
Integrate LLMs, document understanding models, NLP models, rule systems, and algorithm components into a unified pipeline with clear input/output contracts, task orchestration, version governance, and failure handling.
Own engineering capabilities for AI data processing and knowledge services: APIs, async tasks, queues, caching, retry and graceful degradation, human review, canary releases, rollbacks, fault recovery, and capacity governance.
Establish a quality system for AI data processing, knowledge, and RAG—measuring parsing accuracy, retrieval coverage, citation completeness, staleness, latency, stability, and cost, with tiered root-cause analysis.
Collaborate with data, algorithm, product, and compliance teams to deploy AI data processing and knowledge services reliably in user-facing equities products and Binance AI.
Master’s degree or above in Computer Science, Software Engineering, AI, or a related field; 5+ years of experience in backend, data platforms, ML engineering, or AI application engineering.
Familiarity with equity markets and the investor research and decision-making workflow; understanding of trading mechanics, market data, fundamentals, corporate actions, and major market events; ability to assess the entities, time sensitivity, sources, and usage boundaries of financial information.
Proficient in Python and at least one of Java or another backend language; solid software engineering, distributed systems, and service interface design skills.
Production experience with LLMs, NLP, or ML systems; ability to explain model invocation, task orchestration, failure recovery, version governance, and online issue resolution.
Production experience with knowledge engineering or RAG systems; familiarity with structured, semi-structured, and unstructured content processing; ability to explain the full pipeline from source ingestion through retrieval, citation, and online feedback.
Familiarity with full-text search, vector search, document storage, and their combinations; understanding of chunking, indexing, filtering, recall, reranking, context assembly, citation, and permission control—not tied to any specific database or framework.
Experience with performance, stability, cost, and observability governance for high-concurrency or large-scale processing systems—beyond model API calls or demo prototypes.
Experience adapting new data sources or content types; ability to distill source-specific logic into reusable processing capabilities.
Originally posted on Himalayas