Build a Phosphate Rail Route Optimization Agent with LlamaIndex
Doubling Saudi Arabia's $1bn phosphate rail track requires continuous alignment and terrain route intelligence. Build an advanced RAG agent using LlamaIndex with property graph indexing to query geotechnical surveys and route alignment specs efficiently.
What you are building
The core problem, expected build, and operating context for this challenge.
Utilize LlamaIndex Property Graphs and structured data connectors to query Saudi Railway Company (SAR) expansion plans and terrain constraints.
How work is evaluated
Evaluates accuracy of track route query results and safety load compliance validation.
Shared data for this challenge
Review public datasets and any private uploads tied to your build.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
axle_load_check
Verify extracted load matches target specification
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
graph_retrieval_faithfulness
Faithfulness of answer derived from Knowledge Graph nodes • target: 0.97 • range: 0.88-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Implement LlamaIndex PropertyGraphIndex for connected railway domain entities
Parse geospatial terrain reports and track-doubling engineering constraints
Build dynamic sub-question query engines to analyze cross-country rail corridor risks
Generate automated compliance reports for railway safety and axle-load specifications
Reference links and supporting material
Geotechnical reports, heavy-haul railway track standards, and corridor alignment specs for Saudi SAR network.
How this agent runs
Evaluates accuracy of track route query results and safety load compliance validation.
Challenge input
JSON query string
LlamaIndex
Enterprise standard for complex graph indexing and structured document parsing
Evaluated output
JSON spec response
- Verify extracted load matches target specification
- Faithfulness of answer derived from Knowledge Graph nodes • target: 0.97 • range: 0.88-1
- Benchmark: GraphRAGBench
- Graph Retrieval Faithfulness target: 0.97
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
View technical recipe
Configured tools
- Llama Index · Required
- RAI · Optional
- Vercel AI Gateway · Optional
- Vercel AI Gateway · Optional
Evaluation contract
- axle_load_check · Weight 1
- graph_retrieval_faithfulness · Weight 1
Recipe state
This is a preview. The configuration can change before the evaluation recipe is locked.