AI Engineer

Job Requirements
2.1 AI Engineers (2 positions)
The AI engineers will design, build, and iterate on AI agent capabilities within an agentic orchestration framework.

Work will span

Agent orchestration and workflow execution - Implementing multi-step agent pipelines with human-in-the-loop approval gates, state persistence, and resumable execution
Document intelligence - Building extraction and classification capabilities for engineering documents (single-line diagrams, bills of material, mechanical drawings, site profiles, tariff schedules) using LLM-based pipelines
Knowledge graph construction and validation - AI-driven generation, enrichment, and deterministic validation of RDF/OWL knowledge graphs representing physical assets and their relationships
Conversational AI - Extending a chat-based interface that allows operators to query operational data, trigger agent workflows, and review AI-generated recommendations
Device and telemetry classification - Agents that identify and classify new devices and telemetry streams appearing on the network, mapping them to canonical ontology types
Observability and evaluation - Instrumenting agents with structured logging, tracing, and quality scoring to support iterative prompt engineering and reliability improvements
Required skills:

• Strong Python development (3+ years production experience)
• Hands-on experience building LLM-powered applications (prompt engineering, RAG, tool use / function calling, structured output)
• Experience with at least one agent framework (LangGraph, Semantic Kernel, AutoGen, CrewAI, or equivalent)
• Familiarity with document extraction pipelines (OCR, PDF parsing, layout-aware models)
• Working knowledge of RDF, OWL, SPARQL, or similar knowledge representation technologies - Experience with Azure AI services (Azure OpenAI, Azure AI Document Intelligence, or Azure AI Foundry)

• Experience with Model Context Protocol (MCP) tool servers - Background in energy, industrial IoT, or SCADA systems - Familiarity with Sparkplug B or MQTT protocol stacks - Experience with evaluation frameworks for LLM applications (e.g., promptfoo, ragas, or custom harnesses)