Federated Learning
Digitial Twin
Demo Console
Landing Site
Blockchain Audit Trail

SentinelMesh

Inspiration

Critical infrastructure such as power grids, water systems, and manufacturing plants still relies on legacy OT/ICS networks that were never designed for modern cyber threats. These systems cannot be patched easily, tolerate zero downtime, and are increasingly targeted by sophisticated attackers. At the same time, the emergence of quantum computing threatens to break traditional cryptography.

SentinelMesh was inspired by a core question:

How do we protect legacy infrastructure without touching it, slowing it down, or breaking it — while also preparing for post-quantum threats?

What It Does

SentinelMesh is an autonomous, quantum-safe cyber defense fabric designed for legacy critical infrastructure. It passively monitors industrial network traffic, detects anomalies in real time using AI, and enables safe, autonomous responses without disrupting operations.

Key Capabilities

Real-time AI-based anomaly detection
Privacy-preserving federated learning across sites
Digital twin–based predictive monitoring
Post-quantum cryptography for future-proof security
Decentralized identity and blockchain-backed audit trails

How We Built It

SentinelMesh follows a modular, distributed architecture designed to balance realism, safety, and demonstrability.

Tech Stack & Implementation

Component	Tools / Frameworks
Backend	FastAPI, PyTorch, Supabase
Frontend	Next.js, Tailwind CSS
Deployment	Vercel, TypeScript

Key Features

Feature	Description
Decentralized Trust & Audit Layer (Blockchain + IPFS)	Smart contracts for automated policy enforcement and immutable, verifiable audit logs
Federated Learning (Flower Framework)	Privacy-preserving collaborative anomaly detection across sites without sharing raw operational data
ML Pipeline (CatBoost / Time-Series Models)	Optimized anomaly detection for industrial OT traffic with categorical protocol support
Digital Twin–Based Anomaly Detection	Real-time deviation detection using predictive baselines and root-cause analysis

Challenges We Ran Into

Designing security for legacy OT/ICS systems that cannot be modified
Balancing near real-time anomaly detection (<50ms) with safety and low false positives
Representing complex distributed system concepts clearly in a prototype
Integrating AI, post-quantum cryptography, and decentralized trust into a coherent design
Translating production-grade security ideas into a demo-friendly experience

Accomplishments We’re Proud Of

Designing a non-intrusive, quantum-safe security architecture for legacy systems
Demonstrating federated learning for cross-site security without sharing raw operational data
Building a clear, interactive prototype that visualizes complex industrial security workflows
Addressing real-world OT security constraints instead of idealized environments
Creating a scalable design applicable across energy, water, and transportation sectors

What We Learned

Critical infrastructure security requires systems thinking, not isolated tools
AI is most effective when combined with trust, cryptography, and governance
Privacy-preserving collaboration is essential for cross-organization defense
Preparing for quantum threats must begin before quantum computers arrive
Clear visualization builds trust in autonomous security systems

What’s Next for SentinelMesh

Adaptive, AI-Driven Response Policies

Introduce reinforcement learning and adaptive control to autonomously optimize mitigation strategies over time, reducing reliance on human intervention while maintaining operational safety.

Hardware-Accelerated Post-Quantum Cryptography

Integrate FPGA/ASIC-accelerated PQC modules to support high-throughput, real-time encryption for industrial networks without performance impact.

End-to-End Digital Twin Ecosystem

Build multi-site, multi-domain digital twins to simulate full industrial operations, predict cascading failures, and visualize root causes for anomalous events.

Autonomous Threat Simulation & Red-Teaming

Implement adversarial GAN-based simulations to continuously test system resilience against emerging attack vectors, including replay attacks, protocol exploits, and quantum-enabled threats.