Skip to main content
Back to challenges
Challenge

Industrial AI Readiness Compliance Auditor with Google ADK and Daytona

Based on the 'Singapore Launches Industrial AI Readiness Framework' (July 2026), this challenge requires you to build an autonomous compliance auditor. The agent must assess industrial site readiness by analyzing multimodal data—including technical SOPs, shop-floor imagery, and sensor logs. You will use the Google ADK (Agent Development Kit) to build the core agent, leveraging Gemini's multimodal reasoning capabilities to identify gaps between actual operations and the Singapore AI Readiness standards. To ensure the agent operates in a consistent and secure environment, you will deploy it using Daytona, which will manage the agent's development and execution workspace, ensuring that all dependencies and data-processing tools are isolated and reproducible.

Workflow AutomationHosted 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 multimodal auditor using Google ADK and Gemini to evaluate industrial AI readiness, executed within a Daytona-managed sandbox.

Delivery guide

How work is evaluated

Evaluation

Evaluation is based on the agent's ability to accurately classify readiness levels from a mix of imagery and text, and the successful deployment of the workspace via Daytona.

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

Multimodal Input Handling

Verifies the agent correctly processes both the image and text inputs.

Binary check

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

Dimension 2

Daytona Lifecycle Test

Ensures the Daytona workspace was spun up and torn down correctly.

Binary check

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

Dimension 3

Classification Accuracy

Percentage of correctly identified readiness levels. • 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

  • Implement multimodal reasoning using Google ADK and Gemini

  • Orchestrate isolated agent environments using Daytona

  • Translate high-level government frameworks into technical evaluation heuristics

  • Build a data-flywheel for industrial compliance assessments

Resources and assets

Reference links and supporting material

Dataset notes

Dataset consists of 100 industrial audit scenarios, each with one shop-floor image (synthetic) and one corresponding SOP text snippet.

Challenge setup

Evaluation is based on the agent's ability to accurately classify readiness levels from a mix of imagery and text, and the successful deployment of the workspace via Daytona.

Challenge input

Binary image file and a text-based SOP document.

Google ADK

Provides the primary interface for Gemini multimodal interactions.

Daytona

Ensures isolated and reproducible execution of the compliance agent.

Evaluated output

JSON compliance report.

Evaluation checks
  • Verifies the agent correctly processes both the image and text inputs.
  • Ensures the Daytona workspace was spun up and torn down correctly.
  • Percentage of correctly identified readiness levels. • target: 0.9 • range: 0-1
Configured evaluation evidence
  • Benchmark: Smart-Nation-Audit-2026
  • Classification Accuracy target: 0.9
  • 1 public reference case
Configured runtime
  • Docker 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

Environment
  • Daytona · Optional
Action Space
  • Google ADK · Required
  • Daytona · Optional
  • Google · Optional

Evaluation contract

  • Multimodal Input Handling · Weight 1
  • Daytona Lifecycle Test · Weight 1
  • Classification Accuracy · Weight 1

Metrics

  • Classification Accuracy · Target 0.9

Benchmarks

  • Smart-Nation-Audit-2026
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Binary image file and a text-based SOP document.

    Output: JSON compliance report.

    Execution harness: Docker

Evaluation tools

  • Google ADK · Agent Framework
  • Daytona · Agent Environment

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Industrial AI Readiness Compliance Auditor with Google ADK and Daytona