Hi There!
I am Parth Kothari, a person travelling through the various experience of life, hand-in-hand with my cutie wife.
On the professional front, I work in Google Zurich in the CoreML Frontiers team. Our team is at the intersection of research and productionization, wherein we take inspiration from the latest developments in the LLM Agents research and consequently build infrastructure to support the integration of agents into various products. As part of this setup, I actively contribute to several open-source core initiatives, including Cloud’s Agent Development Kit (ADK), DeepMind’s OneTwo framework, and the GenAI processors library.
Thanks to my team’s unique positioning between research and product, I enjoy consulting internal product teams to guide them on effectively incorporating LLM agents into their workflows.
My work at Google builds on a deep learning background shaped by a PhD at the Visual Intelligence for Transport (VITA) lab at EPFL, under the supervision of Prof. Alexandre Alahi in socially aware motion forecasting for autonomous vehicles. I got my undergraduate degree in Electrical Engineering from (IIT Bombay).
Updates
| Oct 2022: | Started working at Google, Zurich! |
| Oct 2022: | Successfully defended my PhD! |
| Oct 2022: | new paper on motion style transfer for trajectory forecasting accepted to CoRL 2022. |
| Sep 2021: | new paper introducing an open-source gym-compatible environment for autonomous driving accepted to ML4AD Workshop at NeurIPS 2021. |
| Sep 2021: | new paper on test-time adaptation of machine learning models accepted to NeurIPS 2021. |
| Jun 2021: | started research internship at Level 5 Woven Planet (previously Lyft). |
| Apr 2021: | gave guest lectures on modelling social interactions and modelling multimodal distributions for human trajectory forecasting for the Deep Learning for Autonomous Vehicles course. |
| Mar 2021: | new paper on interpretable anchor-based method for human trajectory forecasting in crowds accepted to CVPR 2021. |
| Feb 2021: | new paper on modelling human social interactions in crowds and describing our TrajNet++ benchmark to appear in IEEE Transactions on Intelligent Transportation Systems. [Code] |
| Nov 2019: | new paper on collaborative sampling from generative adversarial networks is accepted to AAAI 2020. [Code] |
| Nov 2019: | cleared the EPFL PhD Candidancy Exam. |
| Aug 2018: | graduated from IIT Bombay, with Institute Rank 7. |
| Jun 2018: | secured 3rd position in the iFood Challenge at CVPR 2018. [Code] |
| Apr 2018: | I will be joining VITA Lab as a PhD candidate, under supervision of Prof. Alexandre Alahi, in Fall 2018 |