AI Engineer – GraphRAG & MCP Tooling Development

Description

We are looking for a talented, highly autonomous, and self-driven AI Engineer to take hands-on ownership of our GraphRAG (Graph Retrieval-Augmented Generation) systems and contribute directly to the evolution of TigerGraph’s MCP (Model Context Protocol) tooling framework. This role spans deep AI/LLM integration, graph query pipelines, and developer tooling—helping build a cutting-edge platform that blends graph intelligence with generative AI.

This is a role for a true builder who thrives in ambiguity. You won’t just be handed a neatly packaged list of Jira tickets; you will be expected to explore the bleeding edge of GraphRAG, identify what our framework needs next, and write the code to make it happen. If you are deeply curious, thoughtful in your architectural decisions, and motivated by independently solving complex engineering problems from conceptualization to deployment, you will thrive here.

Responsibilities

● Act as a self-directed engineer: Navigate ambiguous technical challenges, make sound architectural trade-offs, and drive the execution of GraphRAG systems using TigerGraph’s MCP framework without needing a step-by-step roadmap.

● Proactively design and develop: Identify gaps and build essential MCP tools and components from the ground up, including orchestration logic, agentic-AI workflows, LLM interface layers, and graph-native operators.

● Architect with foresight: Build elegant, schema-aware integration code between TigerGraph’s GSQL, vector indexing systems, and external LLMs (e.g., OpenAI, Gemini, LLaMA), anticipating future platform scale.

● Build for developers: Thoughtfully engineer reusable modules, prompts, and components for cognitive agents (e.g., GraphRAG agents, schema routers, grounded QA evaluators) with a strong focus on developer experience and API design.

● Collaborate as a technical partner: Work closely alongside TigerGraph’s platform, AI research, and product teams to shape the MCP engineering roadmap, bringing your own solutions to usability and scalability hurdles.

● Ensure engineering excellence: Independently write test suites and benchmark the performance of GraphRAG systems for hallucination, groundedness, latency, and answer usefulness, holding your own code to the highest standard.

● Empower the community: Write highly readable internal documentation and clear SDKs to improve the developer usability of MCP.

Requirements

Qualifications
Required:

● High Agency & Self-Drive: A proven track record of taking vague technical concepts, figuring out the optimal engineering path, and writing production-ready code without requiring heavy hand-holding or day-to-day micro-direction.

● Product-Minded Engineer: You don't just write scripts; you think deeply about the "why" behind the feature and care immensely about how other developers will interact with your code.

● Strong programming skills in Python; deep hands-on experience building LLM orchestration tools, agent systems, or AI SDKs.

● Hands-on experience with TigerGraph (GSQL queries, RESTPP, schema modeling).

● Familiarity with Graph-based retrieval-augmented generation (GraphRAG) architectures and their application in real-world AI systems.

● Experience using or actively contributing to frameworks like LangChain, LangGraph, or similar agent-based LLM tools and prompt templating.

● Understanding of vector indexing and similarity search; familiar with modern vector stores (e.g., FAISS, Milvus).

● Ability to design exceptionally usable internal tools for developers or data scientists.

Preferred:

● Prior experience developing tools, platforms, or APIs used by other AI engineers or ML practitioners.

● Background in knowledge graphs, graph neural networks, or knowledge-based QA systems.

● Familiarity with Docker/Kubernetes, FastAPI, and distributed compute systems.

● Contributions to open-source projects in the graph, ML, or LLM domains.