Satyandra Guthula
Computer Science PhD student
I’m a Ph.D. student at the University of California, Santa Barbara, working under the supervision of Prof. Arpit Gupta in the Systems and Networking Lab.
Currently, my research focuses on the convergence of networking, security, and machine learning. More specifically, I am dedicated to crafting machine learning-based solutions for challenges in networking and security. My aim is to develop resilient representations capable of generalizing across diverse network conditions and catering to various downstream tasks.
Presently, I'm heading the netFound initiative at UCSB. Our goal is to create foundational models for networking data utilizing self-supervised learning methods. This entails leveraging plentiful unlabeled network data, gathered passively from production environments through PINOT for task-agnostic pre-training. Additionally, we utilize smaller-scale labeled network data, actively collected using PINOT and netUnicorn, for task-specific fine-tuning.
Research Interests
My interests broadly span the following areas:
-
Foundation models for networking and distributed systems
-
Reinforcement learning for QoE estimation
-
Agent based modeling of computer systems and networks
​
My hobby interests are Foundation models in communications and story writing
Education
-
Ph.D. in Computer Science
-
University of California, Santa Barbara 2022 - now, with Prof. Arpit Gupta
-
-
MTech in Computer Science
-
Indian Institute ot Technology Kanpur 2016-18, with Prof. Harish Karnick and Prof. Sunil Simon
-
-
BE in Computer Engineering
-
Mumbai University 2012-16
-
Work Experience
Between 2018 and 2022, I contributed to VMware as a member of the vRNI team, where our primary focus was the development of tools for monitoring data centers, providing a comprehensive overview of various components within these environments.
​
In my role, I centered on the collection of data from diverse network devices, including switches, firewalls, and routers. My responsibilities extended to integrating routing information, enabling analytics and visualizations specific to these network components.
​
Beyond these duties, I actively participated in several research-based projects within the team, which were patented. Specifically, I worked on utilizing network reachability graphs to cluster VMs with similar configurations as part of an application. Additionally, I delved into the application of reinforcement learning-based controllers to optimize device polling frequency, preventing unnecessary cycles on physically unchanged devices.