About this role
Sole owner of the agentic stack for Patsnap's Materials team, responsible for designing, building, and productionizing LLM-powered agents, MCPs, memory systems and evaluation frameworks for materials science search, Q&A and information extraction. This is an in-office position based in the Singapore office, working directly with product managers, domain experts and the platform team.
Skills for this role
Large Language Models (LLMs)Agentic systemsMulti-step reasoningTool orchestrationGuardrailsMemory systemsModel Context Protocol (MCP)Multi-agent systemsEvaluation frameworkspromptfooBraintrustLangSmithDeepEvalOpenTelemetryArize PhoenixLangfuseDatadogPythonProduction reliabilityObservabilityVector databasesKeyword searchKnowledge graphsRerankingHybrid retrievalInformation extractionStructured information extractionMaterials scienceChemistryPatent/IP