Engineer - AI

Responsibilities

• ● We are seeking a Knowledge Engineer to support the development of MAV for structured ingestion, reasoning, and retrieval of complex mining and asset information. The successful candidate will design and implement end-to-end knowledge extraction and ingestion pipelines, transforming unstructured and semi-structured technical documents (PDFs, reports, drawings) into RDF-based knowledge graphs, and enabling graph-native retrieval and reasoning workflows (GraphRAG). This role sits at the intersection of knowledge engineering, AI-driven extraction (LLM + vision), ontology design, and graph analytics.
• ● Responsibilities
• ● Knowledge Extraction & Ingestion
• ● Design and implement automated extraction pipelines for technical tailings storage facility (TSF) documents (PDFs, scanned reports, tables, figures).
• ● Apply LLM-based and vision-based extraction techniques (OCR, layout understanding, multimodal models) to identify entities, attributes, relationships, and evidence.
• ● Develop validation and normalization logic to ensure extracted knowledge meets quality and consistency requirements.
• ● Knowledge Schema & Ontology Design
• ● Design and evolve domain ontologies and knowledge schemas to support structured storage of TSF, risk, asset, and operational data.
• ● Implement schemas using RDF/OWL, including classes, properties, constraints, and semantic relationships.
• ● Align schemas with industry standards and internal MAV data models.
• ● Knowledge Graph Development
• ● Build and manage RDF-based knowledge graphs in graph repositories (e.g., GraphDB, RDFox, Neptune, or equivalent).
• ● Implement ingestion workflows that map extracted content into graph structures with traceability to source documents.
• ● Support versioning, provenance, and evidence linking within the knowledge graph.
• ● Retrieval & GraphRAG Pipelines
• ● Design and implement graph-native retrieval pipelines, combining SPARQL queries, reasoning, and embeddings where appropriate.
• ● Develop GraphRAG architectures that leverage structured graph context rather than flat text retrieval.
• ● Enable natural-language querying over the knowledge graph for downstream AI assistants and analytics tools.
• ● Collaboration & Integration
• ● Work closely with development team, domain experts, and AI engineers to refine extraction logic and schema requirements.