We are seeking a highly skilled and experienced Python Backend Engineer to join our Materials Science AI team. You will take full ownership of the AI Agent architecture, specifically focusing on the development of formulation and search agents. You will design the "brain" of our systems—orchestrating complex workflows between LLMs, internal skill modules, and scientific search APIs. You will collaborate with cross-functional stakeholders, mentor junior engineers, and drive innovation in our Agentic backend services while ensuring a seamless integration with our Vue-based frontend.
Responsibilities
- Design and implement robust backend solutions in Python using a three-layer architecture (Main Agent, Skills, and Search APIs).
- Own the entire development cycle of Formulation Agents, from structured data extraction (patents/literature) to mechanistic reasoning and "functional blueprint" generation.
- Address Out-of-Distribution (OOD) issues, reduce hallucinations through expert knowledge engineering, and ensure high-performance execution of complex agentic chains.
- Develop and maintain modular "Skills" and agentic search tools, ensuring they are optimized for complex scientific querying.
- Work closely with Material & Formulation Experts, Frontend Engineers, and Product Managers to align product roadmaps.
- Diagnose issues in agent reasoning or data pipelines, implementing long-term fixes to prevent regressions in AI performance.
Qualifications
Minimum Qualifications
- Bachelor’s or Master’s degree in Computer Science, Software Engineering, or a related field.
- Strong understanding of AI Orchestration frameworks (e.g., LangChain, LangGraph, or CrewAI) and LLM API integration.
- Basic proficiency in Vue.js or modern frontend frameworks to facilitate full-stack collaboration and UI-to-Agent state management.
- Familiarity with CI/CD pipelines and containerization technologies (e.g., Docker, Kubernetes).
- Familiarity with SQL/NoSQL databases.
-
Demonstrated ability to independently lead AI-centric projects from concept to deployment.
Preferred Qualifications
- Familiarity with the Model Context Protocol (MCP) and building modular "Skill" tools for AI agents.
- Familiarity with Agent technical stack, e.g., agentic search, context engineering, etc.
- Proficient in Python, and Familarity with vibe coding, e.g., Claude code.
- Strong communication skills and the ability to collaborate effectively with a multidisciplinary team.