Sandeep Dommari

I authored original research titled "AI-Based Ensemble Learning Approach for Intrusion Detection in Smart Grid Environments," addressing critical vulnerabilities in modern energy infrastructure. I designed a novel Intrusion Detection System (IDS) using an ensemble learning architecture that integrates Random Forest, Gradient Boosting, and XGBoost algorithms to identify sophisticated cyber threats like DoS and data injection attacks. Testing on benchmark datasets (NSL-KDD, CICIDS2017) demonstrated that my model achieves 99.8% detection accuracy with a minimal 0.2% false positive rate. This work significantly advances the field of cyber-physical security by providing a low-latency, scalable solution capable of running on edge devices. It directly strengthens the resilience of national smart grids, ensuring real-time defense against malicious activities that could disrupt essential power distribution services.

Unlike traditional cyber threats that rely on static scripts or human direction, Agentic AI operates as autonomous, goal-driven entities. These agents can function individually or coordinate as swarms to dynamically discover network topography, exploit vulnerabilities, and adapt to enterprise environments in real-time. They automate critical offensive phases such as reconnaissance, lateral movement, and decision-making, executing complex campaigns at machine speed without requiring constant command-and-control communication. The rise of these autonomous swarms renders traditional perimeter-based and rule-based security models insufficient, as these defenses cannot account for adversaries that reason and pivot autonomously. The primary impact on the enterprise is a forced paradigm shift away from static defenses toward identity-centric and behavior-based detection. Security leaders must now implement adaptive controls and Zero Trust architectures capable of matching the speed and adapt

https://www.bintelligence.com/judge/sandeep-dommari Customer Service Awards Cybersecurity Awards Sales and Marketing Awards Sustainability_Awards Stratus Awards for Cloud Computing Herizon Awards BIG Awards for Business 2026 BIG Innovation Awards Selected as a Judge for the Business Intelligence Group Awards (link: https://www.bintelligence.com/judge/sandeep-dommari), reviewing nominations across categories including Cybersecurity, Customer Service, Sales & Marketing, Sustainability, and Stratus Awards for Cloud Computing, Herizon Awards, BIG Awards for Business, and 2026 BIG Innovation Awards. In this role, I evaluated global submissions on technology innovation, with a focus on how organizations leverage Microsoft Azure, Entra ID, DevOps, and cloud-native security to drive transformation and resilience. By providing structured feedback, I helped highlight industry best practices and recognized excellence in secure, Microsoft-driven digital adoption across enterprises.

I served as a keynote speaker at an international conference focused on AI, cybersecurity, and emerging trends in computer science, where I presented on "AI-Augmented Cyber Defense: Reimagining Security in the Age of Intelligent Threats." My session explored how advancements in machine learning and generative AI are reshaping both cyber threats and defense strategies. I discussed real world use cases, emerging attack patterns, and practical approaches for integrating AI into modern security architectures. The keynote helped practitioners, researchers, and leaders better understand how to leverage AI responsibly to strengthen cyber resilience, protect critical systems, and prepare for the future of intelligent, adaptive threats across industries.

The blog details Neuro-symbolic AI is a hybrid architectural approach that fuses neural networks (which excel at pattern recognition and processing unstructured data) with symbolic AI (which utilizes logic, rules, and knowledge graphs). In this model, neural modules handle "perception" by processing raw telemetry, while symbolic modules handle "reasoning" by applying logical constraints and explainable rules to those outputs, creating a system that can both learn from data and reason about it. This approach fundamentally changes cybersecurity defenses by solving the "black box" problem of traditional deep learning. Its major impact is the delivery of explainable decisions (telling analysts why a threat was flagged) and increased adversarial robustness (using logical guardrails to block attacks that trick pure neural models). This transitions security from simple pattern matching to context-aware reasoning, enabling defenses that are resilient against sophisticated, evolving attacks.