Full Stack AI Solution Engineer
About the role
This role involves enterprise-level LLM, RAG and Agent system engineering, design and optimization. You will undertake research and engineering implementation of complex AI architectures, build hybrid retrieval systems, design business Agent orchestration platforms, and establish systematic prompt engineering and observability frameworks. The position requires independent architecture design, performance tuning and online iteration of large-scale AI applications across multiple technical domains.
Key responsibilities
Undertake research and engineering implementation of enterprise-level RAG systems, build hybrid retrieval architecture integrating keyword, vector and graph retrieval, realize adaptive document chunking and reranking model fine-tuning, land full-link hallucination suppression strategies, and continuously improve knowledge retrieval accuracy and LLM answer robustness
Design and build complex business Agent orchestration platforms, implement multi-tool chained calling, hierarchical memory management, dynamic prompt routing and automatic Agent failure retry mechanisms, and standardize human-machine collaborative workflow for complex industrial and business scenarios
Build standardized prompt engineering systems, complete NL2SQL solution design and tuning, optimize conversational data analysis interaction logic and user intent recognition capabilities, establish prompt version management, quantitative effect evaluation and multi-turn dialogue intent clarification mechanisms
Build end-to-end observability systems for LLM applications, realize prompt parameter tracking, retrieval log collection and monitoring, model output hallucination detection, and automatic user Q&A effect scoring
Encapsulate reusable AI application basic components including general knowledge base access modules, Agent tool plugin market and unified multi-model invocation gateway. Build A/B testing systems to support gray-scale comparison and iterative optimization of RAG retrieval strategies, prompt templates and Agent business logic
Optimize operating costs of large-scale AI applications via LLM caching, quantitative inference, vector storage tiered management and idle computing power recycling scheduling. Build AI security and compliance architecture covering prompt injection prevention, knowledge base data desensitization, model access authentication and user dialogue data standardized compliance management
Design cross-terminal unified AI service gateway, unify logging, authentication and rate limiting standards for RAG, Agent and LLM fine-tuning businesses
Promote the upgrade of POC-level AI solutions to enterprise-scale architectures, implement vector database sharding and incremental knowledge synchronization, optimize massive document retrieval performance
Sort out structured workflows for complex industrial Agents, convert technical architectures and indicators into quantifiable business value and high-level reporting metrics
About you
Solid practical experience in end-to-end enterprise-level LLM, RAG and Agent project landing, capable of independent architecture design, performance tuning and online iteration of large-scale AI applications
In-depth understanding of systematic prompt engineering and NL2SQL technology, proficient in multi-turn dialogue context management and intent clarification, with practical experience in prompt version management and effect quantitative evaluation
Familiar with vector database engineering, hybrid retrieval optimization and reranking tuning, able to solve core problems such as insufficient retrieval accuracy, model hallucination and massive data retrieval performance bottlenecks
Proficient in Databricks platform application and MLOps system construction, familiar with data lake + vector database fusion architecture and Delta Lake data governance tuning
Have in-depth practice in AI application cost optimization, security compliance architecture construction and technical debt governance, with enterprise-level high availability, standardization and compliance awareness
Possess excellent business thinking and cross-team communication capabilities, able to translate technical advantages into business value and promote standardized and large-scale landing of AI projects