Senior AI Engineer (Generative AI / LLM Systems)

Quick Overview

• Build production AI systems that reason over industrial knowledge graphs (PlantGraph).
• Work on LLMs, RAG, and agent-based systems solving real-world engineering problems.
• Integrate AI with structured data, diagrams, and operational workflows.
• Own complex, ambiguous problems end-to-end in a high-impact domain.

Description

Build an AI-native platform that helps industrial companies understand and operate complex facilities. PlantGraph is an ontology-driven knowledge graph that models equipment, instrumentation, and process relationships across a facility using engineering diagrams (P&IDs), documentation, and operational data.

Embed AI directly into the system to enable:

• Natural-language interaction with facility data.
• Graph-aware reasoning over engineering systems.
• AI agents that operate across diagrams, documents, and workflows.

This work focuses on making AI reliable, grounded, and usable in real-world engineering environments.

What You’ll Do

Build AI Systems That Reason Over Structured Industrial Data

• Design systems that allow LLMs to interpret and reason over PlantGraph and its underlying ontology.
• Combine graph queries, ontology structures, and engineering data into reliable, explainable outputs.

Create Natural Language Interfaces Over Complex Systems

• Build chat-based experiences that allow users to explore facility systems, navigate diagrams, and query equipment and process relationships through conversation.

Orchestrate AI Across Graphs, Documents, and Workflows

• Develop systems combining graph queries, engineering documentation (P&IDs, procedures, LOTO, and work orders), and real-world operational context.
• Enable accurate, traceable AI outputs.

Enable AI Agents to Safely Interact with the Platform

• Design APIs and tools that allow AI agents to operate on PlantGraph and system capabilities.
• Ensure interactions are observable, reliable, and production-safe.

Productionize AI Systems at Scale

• Turn prototypes into production systems.
• Build scalable APIs and services.
• Optimize performance and cost.
• Develop evaluation, monitoring, and reliability frameworks.

Own Ambiguous, High-Impact Problems

• Work across engineering, ML, and domain teams to define and solve complex problems.
• Identify and address gaps in data, ontology, and system design.

Core Engineering Challenges

• Grounding LLMs in structured graph data.
• Reliable agent workflows across multiple data sources.
• Query optimization across graph, vector, and document systems.
• Ensuring correctness, traceability, and validation in AI outputs.
• Building production-grade AI systems for real-world industrial use.

Required Qualifications

• 5+ years in software engineering, ML engineering, or applied AI.
• Experience building AI systems that combine structured data with LLMs.
• Strong experience with RAG, embeddings, and retrieval systems.
• Experience building production AI systems, not just prototypes.
• Strong Python and backend engineering experience.
• Experience designing scalable APIs and services.
• Ability to take ownership of complex, ambiguous problems.

Preferred Qualifications

• Experience with LLM agents or tool-based AI systems that interact with external systems through APIs or structured tools, including familiarity with emerging standards such as MCP.
• Knowledge graph or graph database experience.
• Exposure to industrial systems, P&IDs, or engineering workflows.
• Experience with PyTorch or TensorFlow.
• Distributed systems or cloud infrastructure experience.

Tech Stack

• LLMs: OpenAI, open-source models (Hugging Face).
• AI Frameworks: LangChain, LlamaIndex, MCP.
• ML: PyTorch, TensorFlow.
• Data: Vector DBs, FalkorDB (graph), and hybrid retrieval systems supporting PlantGraph and structured reasoning over engineering data.
• Backend: Python services and APIs.
• Frontend: .NET-based applications.

Why This Role

This role focuses on real AI problems rather than generic chatbots or isolated prototypes. You will build AI systems used to operate real-world infrastructure that reason over structured engineering systems, integrate deeply into workflows, and must be correct, explainable, and production-ready.