Vamshi Mohan
Sr Developer @Amrita Center For Cybersecurity Systems & Networks
Signup · Get unlimited contacts
WORK HISTORY
Sr Developer @Amrita Center For Cybersecurity Systems & Networks
1) Converted PyTorch adversarial attacks (SSAH, Patch-Fool, FPR) to TensorFlow with cross-dataset/model supportand Docker deployment, reducing execution time by 80% and boosting GPU utilization by 60%.2) Currently researching to develop custom adversarial attacks and detection techniques for image perturbations3) Developed resource allocation, load balancing and rate-limiting using XGBoost, dynamic scaling using Deep Q-Network and network route selection using Ant Colony Optimization to optimize 5G resources4) Integrated GNN and Bayesian models into a rule-based TDD to enhance attack detection accuracy by 89% on device, router and controller layers across user and network planes5) Implemented 5G slice isolation using NFV and traffic segregation to ensure secure operation6) Built intrusion detection using K-Means and GRU, increasing threat mitigation rates by 85%7) Developed sentiment analysis using TF-IDF with 95% accuracy to detect phishing attacks8) Developing interpretability of State Space Models (Mamba) by analyzing activations, performing ablations and benchmarking against transformers to replicate and explain LLM-like behavior9) Developing indirect prompt injection mitigation in Agentic AI via attack analysis, multi-LLM detection and security layers for intention-based user/system prompt categorization, dynamic prompt summarization and adversarial retraining while reducing user fatigue
EDUCATION
Amrita Vishwa Vidyapeetham
Doctor of Philosophy - PhD, Cyber Security and Machine Learning
Bangalore Institute of Technology
Bachelor of Engineering - BE, Electrical and Electronics Engineering
University of Massachusetts Lowell
Master's degree, Computer Engineering
ABOUT VAMSHI MOHAN
Your go-to expert for cybersecurity and AI innovations! Currently advancing my Ph.D. in…
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.