Skip to main content
Loading navigation
Back to challenges

Build Wastewater Aeration Anomaly Agent with Vercel AI SDK and Gemini 2.5 Flash

Wastewater aeration retrofits often generate noisy PLC telemetry that obscures dissolved oxygen drift and blower failures. Build a real-time monitoring agent using Vercel AI SDK and Gemini 2.5 Flash to detect telemetry anomalies and output structured control mitigation actions. Achieve 92% anomaly detection precision across simulated aeration basin streams.

EngineeringHosted by Vera
Sign in free to start

What you are building

Implement an AI SDK agent utilizing Gemini 2.5 Flash to process time-series PLC telemetry from aeration basins and issue structured control recommendations.

How work is evaluated

Evaluation

Evaluates the AI SDK agent's ability to classify telemetry anomalies and generate valid setpoint adjustments.

Datasets

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

How submissions are scored

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

json_schema_validity

Ensures the response matches the required setpoint mitigation schema.

Binary check

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

Dimension 2

detection_precision

Precision of anomaly classification over test telemetry streams. • target: 0.92 • 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

  • Process high-frequency dissolved oxygen and blower current telemetry streams

  • Implement structured output streaming using Vercel AI SDK and Gemini 2.5 Flash

  • Trigger automated setpoint adjustments when sensor drift exceeds safety operational thresholds

Resources

Dataset notes

Simulated minute-by-minute PLC aeration basin sensor readings including dissolved oxygen, valve positions, airflow, and blower motor currents.

Challenge setup

Evaluates the AI SDK agent's ability to classify telemetry anomalies and generate valid setpoint adjustments.

Challenge input

JSON containing telemetry arrays (DO, air flow, blower current)

Vercel AI SDK

Agent execution and structured streaming in TypeScript

Gemini 2.5 Flash

Fast multimodal and structured extraction model

Evaluated output

JSON with anomaly_detected boolean, severity, and suggested_setpoint

Evaluation checks
  • Ensures the response matches the required setpoint mitigation schema.
  • Precision of anomaly classification over test telemetry streams. • target: 0.92 • range: 0-1
Configured evaluation evidence
  • Benchmark: Industrial Sensor Anomaly Evaluation
  • Detection Precision target: 0.92
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Vercel AI SDK · Required
  • Google Gemini · Optional
Policy Serving
  • Gemini 2.5 Flash · Optional
Orchestration
  • Vercel AI SDK · Required

Evaluation contract

  • json_schema_validity · Weight 1
  • detection_precision · Weight 1

Metrics

  • detection_precision · Target 0.92

Benchmarks

  • Industrial Sensor Anomaly Evaluation
  • energy

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing telemetry arrays (DO, air flow, blower current)

    Output: JSON with anomaly_detected boolean, severity, and suggested_setpoint

    Execution harness: JavaScript

Evaluation tools

  • Vercel AI SDK · framework
  • Gemini 2.5 Flash · llm

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build Wastewater Aeration Anomaly Agent with Vercel AI SDK and Gemini 2.5 Flash