Back to challenges
Challenge

Google ADK & Modal Anti-Scam Verification Service

Deploy a Google ADK agent running on serverless Modal infrastructure to intercept and analyze multimodal phishing scams and fraudulent SMS vectors targeting Singapore residents. Achieve sub-500ms response time with 99% phishing identification accuracy.

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.

Utilize Google ADK and Gemini multimodal vision with serverless Modal container execution to rapidly classify phishing vectors.

Delivery guide

How work is evaluated

Evaluation

Evaluates detection precision of scam messages and container response latency on Modal execution.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

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

Sub-500ms Latency Test

Verifies that total processing time is under 500ms

Binary check

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

Dimension 2

scam_detection_accuracy

Accuracy in correctly identifying phishing scams • target: 0.99 • 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 multi-modal verification pipelines with Google Agent Development Kit (ADK)

  • Deploy serverless, ultra-low latency agent workers on Modal cloud infrastructure

  • Analyze SMS text and screenshot artifacts for impersonation and phishing signals

  • Integrate Anti-Scam Command Centre (ASCC) registry feeds for instant matching

Resources and assets

Reference links and supporting material

Dataset notes

Dataset of Singapore-specific phishing SMS templates, spoofed URLs, and legitimate institutional alerts.

How this agent runs

Evaluates detection precision of scam messages and container response latency on Modal execution.

Challenge input

JSON containing message text and image URL

Google ADK

Google's Agent Development Kit for Gemini multimodal orchestration

Modal

Serverless execution environment for low-latency agent functions

Evaluated output

JSON containing label, confidence, indicators, and latency_ms

Checks for
  • Verifies that total processing time is under 500ms
  • Accuracy in correctly identifying phishing scams • target: 0.99 • range: 0-1
Proof of success
  • Benchmark: PhishBench / MultiModalAntiScam
  • Scam Detection Accuracy target: 0.99
  • 1 public reference case
Runtime evidence
  • Python execution harness
  • Python sandbox (unavailable on Versalist)
View technical recipe

Configured tools

Environment
  • Modal · Optional
Action Space
  • Google ADK · Required
  • Google · Optional
Policy Serving
  • Modal · Optional

Evaluation contract

  • Sub-500ms Latency Test · Weight 1
  • scam_detection_accuracy · Weight 1

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Frequently Asked Questions about Google ADK & Modal Anti-Scam Verification Service