Skip to main content
Back to challenges
Challenge

Build a Voice-Enabled Substation Procurement RAG Agent with LlamaIndex and LiveKit

High-voltage substation EPC bidding involves complex technical specifications and rapid query responses for site engineers. Build an interactive voice and document query agent using LlamaIndex and LiveKit that synthesizes technical schematics and procurement constraints with under 1.5 seconds latency.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Develop a voice-driven technical document RAG agent using LlamaIndex indexers and LiveKit real-time voice sessions.

Delivery guide

How work is evaluated

Evaluation

Evaluates accuracy of answers extracted from substation contract PDFs and latency over LiveKit voice channel.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

latency_test

Ensures query response time is under 1500ms

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

retrieval_precision

Cosine similarity of generated response with technical reference specification • target: 0.9 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Build high-performance vector indices for engineering specifications using LlamaIndex

  • Stream natural voice queries and synthesized answers using LiveKit WebRTC SDK

  • Implement dynamic vector routing for Riyadh Expo 380kV substation technical codes

  • Measure and optimize end-to-end voice query latency below 1500ms

Resources and assets

Reference links and supporting material

Dataset notes

Indexed specification documents covering 380kV electrical substation engineering standards, procurement BOMs, and Expo site layout specs.

Challenge setup

Evaluates accuracy of answers extracted from substation contract PDFs and latency over LiveKit voice channel.

Challenge input

Audio chunk or audio transcript string query

LlamaIndex

Vector indexing and document retrieval

LiveKit

Real-time audio session management

Evaluated output

JSON containing string response, source_node_id, latency_ms

Evaluation checks
  • Ensures query response time is under 1500ms
  • Cosine similarity of generated response with technical reference specification • target: 0.9 • range: 0-1
Configured evaluation evidence
  • Benchmark: RAG QA Standard Bench
  • Retrieval Precision target: 0.9
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • LiveKit · Required
  • Llama Index · Optional
  • Exa · Optional

Evaluation contract

  • latency_test · Weight 1
  • retrieval_precision · Weight 1

Metrics

  • retrieval_precision · Target 0.9

Benchmarks

  • RAG QA Standard Bench
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Audio chunk or audio transcript string query

    Output: JSON containing string response, source_node_id, latency_ms

    Execution harness: Python

Evaluation tools

  • LlamaIndex · RAG Framework
  • LiveKit · Voice Streaming Interface

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build a Voice-Enabled Substation Procurement RAG Agent with LlamaIndex and LiveKit