MCP Edge AI for Satellite Data Fusion
Develop an advanced MCP-enabled edge AI system simulating a satellite-based sensor platform. This challenge focuses on fusing real-time sensor data (simulated telemetry and imagery captions) with ground-based contextual information (news articles, geopolitical databases) to identify anomalies or critical events. Participants will design a hybrid reasoning architecture, leveraging Gemini 3 Pro's multimodal capabilities for instant classification and deep analysis, while integrating it with Semantic Kernel for orchestrating complex skills and LlamaIndex for advanced RAG over diverse data sources. The goal is to demonstrate efficient and secure data synthesis at the edge, communicating results to a ground station via a robust MCP. This project highlights the complexities of operating LLMs in constrained environments and the power of intelligent agents for critical infrastructure monitoring.
What you are building
The core problem, expected build, and operating context for this challenge.
Build an MCP-enabled edge AI system using Semantic Kernel and LlamaIndex to fuse simulated satellite data with ground intelligence via Gemini 3 Pro's hybrid reasoning.
How work is evaluated
The evaluation will assess the system's ability to accurately fuse disparate data types, identify simulated anomalies, and communicate results efficiently via MCP. Efficiency of LLM usage under adaptive budgets will also be scored.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Develop an adaptive thinking budget mechanism for efficient LLM operation in constrained environments.
Implement a hybrid reasoning system for real-time data classification and deep contextual analysis.
Orchestrate multimodal data fusion using Semantic Kernel and LlamaIndex with Gemini 2.5 Pro.
Design a robust MCP for secure edge-to-ground communication and tool integration.
Reference links and supporting material
Sample data for 1 tasks
Challenge setup
The evaluation will assess the system's ability to accurately fuse disparate data types, identify simulated anomalies, and communicate results efficiently via MCP. Efficiency of LLM usage under adaptive budgets will a...
Challenge input
{'satellite_telemetry': {'timestamp': '...', 'sensor_readings': {...}, 'location': '...'}, 'imagery_caption': '...', 'ground_news': [{'title': '......
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{'timestamp': '...', 'location': '...', 'anomaly_detected': true, 'anomaly_type': '...', 'confidence': 0.9, 'fused_summary': '...', 'llm_tokens_use...
- Verifies successful and secure transmission of fused data to a simulated ground station via MCP.
- Checks if deep reasoning mode was correctly triggered for complex anomaly detection scenarios.
- AnomalyDetectionAccuracy target: 0.9
- Python execution harness
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
No tool records are attached.
Evaluation contract
- Verifies successful and secure transmission of fused data to a simulated ground station via MCP.
- Checks if deep reasoning mode was correctly triggered for complex anomaly detection scenarios.
Metrics
- AnomalyDetectionAccuracy · Target 0.9
- TokenEfficiency · Target 0.8
Workflow tasks
- Task 1
Input: {'satellite_telemetry': {'timestamp': '...', 'sensor_readings': {...}, 'location': '...'}, 'imagery_caption': '...', 'ground_news': [{'title': '...', 'content': '...'}]}
Output: {'timestamp': '...', 'location': '...', 'anomaly_detected': true, 'anomaly_type': '...', 'confidence': 0.9, 'fused_summary': '...', 'llm_tokens_used': {'instant': 100, 'deep': 500}}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.