Mahesh Arumugam
Biography
Mahesh Arumugam is a software architect passionate about designing, programming, and deploying systems. Currently, Mahesh works in data security and analytics domain. He is a technical professional with over 17 years of diverse and progressively responsible experience in distributed systems, security, infrastructure as code, research and development, team leadership, product development, engineering, problem-solving and mentoring. He has published his research work at various conferences and journals. In addition, he has over 15 patents accepted/submitted covering his work in the industry.
Interests
- Data Science
- Machine Learning
- LLM
- Distributed Systems
- Programming
Education
Master's in Information and Data Science (MIDS), 2023
UC Berkeley
PhD in Computer Science and Engineering, 2006
Michigan State University
MS in Computer Science and Engineering, 2003
Michigan State University
BE in Computer Science and Engineering, 2001
College of Engineering, Guindy, Anna University
Experience
Architect
Principal Software Engineer
- Responsible for the architecture and design of Molecula as a Service platform that enables customers to (1) store features in an optimized format, (2) retrieve responses for complex analytical queries quickly, (3) serve features to analytical and ML engines. In this role, contributed to development of control plane for Molecula as a Service that handles management of organizations/tenants, users, roles, deployment of feature store instances, queries to tables on a deployment, etc.
Principal Software Engineer
- Responsible for design and development of connector framework for Tetration. The control plane in- cludes provisioning, configuring, monitoring, and troubleshooting connectors. And, designed and developed various connectors that collect network telemetry from routers, switches (NetFlow / IPFIX), middle- boxes, and VPN endpoints. In addition, designed and developed annotation framework for LDAP annotations on flow telemetry.
- Designed and developed a full-stack cloud security posture management system for AWS accounts (e.g., CIS AWS benchmark), with periodic cloud posture PDF report generation.
Principal Software Engineer
- Responsible for design and development of a declarative configuration language (StyleBooks) for NetScaler, an application delivery controller. Wrote the compiler to generate Python packages from Stylebooks source files.
- Designed and developed the integration of NetScaler with Cisco SDN frameworks such as Cisco Application Centric Infrastructure (ACI) and Cisco Remote Integrated Services Engine (RISE).
- Designed and developed an eventually consistent key-value store in the data-path of NetScaler to manage session persistence information across of a cluster of NetScalers.
Senior Software Engineer
- Responsible for design and development of control plane software for network protocols such as IP and MPLS, traffic engineering (TE) and Fast Re-Route (FRR) technologies, and access control list for Cisco Edge Router ASR9K. In addition, worked on scale optimizations to improve convergence in routing protocols.
Accomplishments
Data Science Specialization (Johns Hopkins University)
Projects
*
In this project, we investigate the urban land use dataset from UC Merced that consists of 21 land use categories from various urban areas across the United States.
A framework for authenticatin sources and lineages of digital objects.
In this paper, we focus on the task of automated code summarization that translates Python source code into a natural language docstring.
In this project, we seek to address the following question: can weather and airline data collected two hours before the scheduled departure time, predict whether the flight will be delayed 15 minutes or more?
In this project, we seek to understand how vaccination rates are impacting vacation mobility in the Pacific Northwest during summer 2021.
Connectors bring in telemetry and analytics data from various vantage points in a data center.
StyleBooks is a declarative language that allow users to consume NetScaler (now, called Citrix ADC) services in a variety of data center configurations and cloud architectures, providing both configuration simplification and smart operational visibility.
DHT provides a key-value store for NetScaler (now, called Citrix ADC) packet engines to store application state in multiple cores of a node or across multiple nodes/cores in a cluster.
Featured Publications
Publications
Quickly discover relevant content by filtering publications.
(2009).
Stabilizing Interference-Free Slot Assignment for Wireless Mesh Networks.
In Guide to Wireless Mesh Networks.
Popular Topics
causal delivery
collision-free communication
convergecast
data dissemination
Data Science
distributed systems
fast actions
logical timestamps
Machine Learning
MIDS
Professional
read/write model
self-stabilization
sensor networks
SF shared-memory model
shared-memory model
slow actions
time-division multiple access
timeliness
transformations