Nikhil Sallagundla

Drive enterprise AI transformation by architecting and deploying multi-modal AI systems that integrate computer vision (SAM, Grounding-DINO) with generative AI platforms to automate complex E-Commerce workflows, delivering scalable end-to-end solutions that optimize operational efficiency and unlock new revenue streams through intelligent image and video processing pipelines.

Developed a revolutionary LLM benchmarking methodology that evaluates reasoning pathways rather than just final outputs, addressing the critical industry gap where models can arrive at correct answers through flawed logic. Created an automated pipeline that traces how LLMs "think," enabling the identification of brittle reasoning patterns that pose risks in production environments. This paradigm shift reduced benchmarking time from one week to two hours while establishing new evaluation standards that prioritize reasoning robustness—essential for reliable enterprise AI deployment.

Advancing Real-Time DDoS Detection Through Bio-Inspired Optimization

Enhancing cybersecurity defense systems by applying bio-inspired optimization algorithms (Particle Swarm Optimization, Chemical Reaction Optimization, and Hybrid Binary Chemical Reaction Optimization) to improve DDoS attack detection accuracy and response times. Building upon established research (doi.org/10.1016/j.eswa.2023.120404), this work focuses on optimizing machine learning models for faster threat identification and mitigation, addressing the critical need for real-time network security in enterprise environments where milliseconds can determine system integrity.