About
Hi, I’m Sarath, a staff (data) scientist at Salesforce. I became part of the Salesforce family through an acqusisition, Tenyx, where I was a research/founding member building key technologies in LLM training, safety, and alignment.
Previously, I obtained my Ph.D. from the Ming Hsieh Department of Electrical and Computer Engineering at the University of Southern California, where I worked with Prof. Antonio Ortega. During my Ph.D., I focused on non-parametric, neighborhood/graph algorithms for understanding data and machine learning models. For more details, checkout my Thesis, Defense Slides.
Want to connect, collaborate, or chat about grad school, research, etc.? Schedule a meeting here: Google Meet
What’s New
[Jan 2025] Will be speaking at Salesforce TDX 2025 on what makes voice AI different! Reach out if you want to chat/hang out!
[Sept 2024] Tenyx joins Salesforce to built Agentforce’s next generation voice and AI technology - Announcement
[July 2024] Will be attending ICML, 2024 in Vienna to present our work LLM geometry and toxicity. Ping me to chat/hang out!
[July 2024] Paper out on arXiv: Reasoning in Large Language Models: A geometric perspective
[Apr 2024] Work with Pratyusha accepted to Open Journal of Signal Processing. Arxiv version here
[Jan 2024] Leading the work on TenyxChat, a series of language models trained to function as useful assistants. Check it out on HuggingFace
[Dec 2023] Blog post on one of our research at Tenyx: Forgetting and Toxicity in LLMs: A Deep Dive on Fine-Tuning Methods
[Nov 2023] Paper out on arXiv: Characterizing Large Language Model Geometry Solves Toxicity Detection and Generation
Background
Prior to joining USC for my Ph.D., I worked as a software engineer for 2.5 years at KLA-Tencor’s Wafer Inspection Division on defect classification software.
I obtained my Master’s degree with a specialization in computer vision and machine learning from USC. During my study, I had the opportunity to work as part of the computer vision research group at USC Institute for Robotics and Intelligent Systems, where I was advised by Dr. Jongmoo Choi under the guidance of Prof. Gerard Medioni. I was also a part-time software developer at the Laboratory of Neurological Imaging, where I worked on the project Informatics Visualization in NeuroImaging (INVIZIAN) led by Prof. John Van Horn.
I completed my Bachelors degree in Electronics and Communication from National Institute of Technology - Tiruchrappalli, where I worked under the advise of Profs. Deivalakshmi and Palanisamy.