POSTDOCTORAL FELLOW
Khalifa University of Science and Technology
Total years of experience :15 years, 6 Months
⢠Apply homomorphic and post-quantum cryptography techniques, e.g. NTRU, McEliece, HQC, etc., to encrypt machine learning models.
⢠Convert encrypted models from python to Verilog design and implement them on FPGAs.
⢠Develop a neural network-based progressive learning model by parallel hyper-parameter tuning.
⢠Use Bayesian and genetic algorithms under the compute budget constraint to build a real-time network anomaly detection model.
⢠Develop a cloud workload for performance analysis of progressive learning model in a container platform.
⢠Cryptomining malware detection using systems calls and machine learning techniques.
⢠Build a dynamic compact neural network model for progressive learning by frequent pruning and expansion.
⢠Dynamic auto-selection and auto-tuning of machine learning models for network anomaly detection.
⢠Containerize the machine learning model under Docker and Kubernetes platform.
⢠Causal analysis of network link between microservices in a multi-tier cloud application using Istio service mesh.
⢠Performance analysis of IBMâs cloud bench tool (cbtool) on Amazon Web Services (AWS).
⢠Autoscaling of AWS EC2 virtual machines using cbtools and k-nearest neighbor model.
.
⢠Class representative in B.tech 3í µí±í µí±
year at Sardar Vallabhbhai National Institute of Technology between
⢠Application developer for e-commerce company House of Fraser in C# .NET.
⢠Manage SQL database and backend IT support for www.houseoffraser.co.uk.
Teaching and Leadership
⢠Lab instructor for an undergraduate course âData Analyticsâ at Khalifa University between Jan to June 2022.
⢠Teaching assistant for a graduate course âSustainable Energyâ at Masdar Institute for five semesters between
⢠Thesis: Machine Learning Methods for the Automated Management of Cloud Computing Workloads. â¢
⢠Thesis: Workload-Driven Power and Thermal Management in Multicore Processors â¢
⢠Thesis: Workload-Driven Power and Thermal Management in Multicore Processors â¢