Skip to main content
Back to challenges
Challenge

Type-Safe Sodium-Ion BESS Monitoring Agent with Pydantic AI and Groq

CATL's recently announced Tener container uses sodium-ion technology, which requires specific thermal and voltage monitoring parameters different from traditional Li-ion. In this challenge, you will build a production-ready monitoring and predictive maintenance agent using the Pydantic AI framework. Pydantic AI's type-safety ensures that every telemetry reading (Voltage, Temp, State of Charge) is validated against strict physical schemas before being processed by the LLM. Using Groq Cloud for ultra-low latency inference, your agent will process high-frequency sensor data to detect early signs of thermal runaway or cell imbalance. The system must use Pydantic's structured outputs to provide clear, actionable maintenance tickets for field technicians. You will implement a dependency injection system in Pydantic AI to switch between live sensor feeds and historical datasets, ensuring a robust testing environment for this critical energy resilience technology.

Machine LearningHosted 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 type-safe, low-latency monitoring agent for CATL sodium-ion containers using Pydantic AI and Groq Cloud.

Delivery guide

How work is evaluated

Evaluation

Evaluation measures the latency of the agent's response and the precision of its anomaly detection against a ground-truth set of faulty battery telemetry.

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

Schema Validation

Agent must reject invalid input types (e.g., string voltage) before calling the LLM.

Binary check

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

Dimension 2

F1-Score Anomaly Detection

Accuracy in identifying thermal events in the test set. • target: 0.98 • 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 Pydantic AI agents with strictly typed result schemas for industrial telemetry

  • Utilize Groq Cloud's LPU inference for real-time anomaly detection (sub-100ms)

  • Design Pydantic models for sodium-ion battery specific characteristics (e.g., lower voltage plateaus)

  • Master dependency injection in Pydantic AI for mocking sensor data streams

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic high-frequency sensor telemetry for sodium-ion battery cells, including normal operating ranges and simulated failure modes (short circuit, thermal runaway).

Challenge setup

Evaluation measures the latency of the agent's response and the precision of its anomaly detection against a ground-truth set of faulty battery telemetry.

Challenge input

Pydantic Model (TelemetryData)

Pydantic AI

Ensures industrial sensor data matches strict physical schemas.

Groq Cloud

Provides the sub-100ms latency required for real-time BESS safety shutdowns.

Evaluated output

Pydantic Model (AgentResponse)

Evaluation checks
  • Agent must reject invalid input types (e.g., string voltage) before calling the LLM.
  • Accuracy in identifying thermal events in the test set. • target: 0.98 • range: 0-1
Configured evaluation evidence
  • Benchmark: BMSControlBench
  • F1 Score Anomaly Detection target: 0.98
  • 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
  • Pydantic AI · Required
  • Kombai · Optional
Policy Serving
  • Groq Cloud · Optional
Orchestration
  • Pydantic AI · Required

Evaluation contract

  • Schema Validation · Weight 1
  • F1-Score Anomaly Detection · Weight 1

Metrics

  • F1-Score Anomaly Detection · Target 0.98

Benchmarks

  • BMSControlBench
  • robotics

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Pydantic Model (TelemetryData)

    Output: Pydantic Model (AgentResponse)

    Execution harness: Python

Evaluation tools

  • Pydantic AI · Type-Safe Agent Framework
  • Groq Cloud · LPU Inference

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Type-Safe Sodium-Ion BESS Monitoring Agent with Pydantic AI and Groq