Projects / Overview
SMR Control Room Decision Support System
Architected a safety-critical AI solution architecture for a Small Modular Reactor (SMR) control room, combining real-time digital twin analytics with human-factors data. Built a dual-agent workflow featuring an AI Safety Supervisor that validates proposed operator actions and provides explainable, step-by-step guidance during critical scenarios.
Overview
Designed and prototyped a safety-critical AI decision support system for Small Modular Reactor (SMR) control room operations. Developed in part for the CI Competition 2026 and grounded in 2 research papers, the platform integrates a 6-subsystem architecture pairing real-time digital twin analytics with dynamic multi-agent AI to monitor plant state, evaluate operator risk, and deliver explainable, step-by-step guidance during critical interventions.
Key Features & Engineering
- 6-Subsystem Architecture: Designed the complete system blueprint based on 2 foundational research papers—defining application components, API interfaces, data pipelines, and safety requirements across all operational modules.
- Competition-Grade Design: Built as part of the CI Competition 2026, tailoring the application architecture to meet rigorous safety, human factors, and operational benchmarks.
- Dual-Agent Safety Workflow: Engineered a multi-agent AI system featuring a dedicated Safety Supervisor agent that validates proposed operator actions against safety protocols before execution.
- Explainable AI (XAI) Guidance: Built conversational reasoning models that translate complex plant telemetry and human-factors data into clear, step-by-step diagnostic insights for operators.
- Scenario-Based Validation: Defined software testing measures to benchmark recovery times and prevent unsafe actions, driving rapid iterative optimization across simulated emergency scenarios.