Inside Plus: Vignesh Subramanian, Senior Research Engineer

Vignesh is a Senior Research Engineer at Plus who has always been fascinated by the possibilities of machine learning and its application to real-world challenges. His journey with Plus began as an intern, where he quickly saw the impact of his work and found a strong sense of purpose in contributing to safer, smarter transportation. After earning his PhD in Electrical Engineering and Computer Science from UC Berkeley, he joined Plus to help advance AI-driven autonomous driving technology.

1. Tell us about your role at Plus.

I focus on designing, implementing, and training machine learning models to solve challenges in the autonomous driving stack. I work closely with the data, simulation, and mapping teams to curate the vast datasets required for training, and I collaborate with the runtime integration team to ensure seamless deployment of the models.

2. What inspired you to pursue a career in autonomous driving technology? What attracted you to join Plus?

My passion for working with machine learning models trained on real-world data, combined with the opportunity to solve practical challenges, inspired me to pursue a career in autonomous driving technology. Initially, for my internship I was particularly drawn to Plus for its unique strategy of applying Level 4 software to different use cases, covering driver-assist and driver-out. Implementing this technology in driver-in scenarios today allows for the rapid deployment of models and the chance to see a positive impact on the road.

The team at Plus brings together expertise from diverse fields and has created a supportive and collaborative environment where you can continuously learn from others and leverage a broad spectrum of knowledge and skills. After experiencing this firsthand and seeing the team’s shared commitment to safer and more energy-efficient mobility, I was eager to join full-time.

3. Can you share a project you worked on that you are particularly proud of?

One project I’m particularly proud of involved working in the generative AI space, focusing on world models for autonomous driving. We utilized self-supervised training techniques that eliminated the need for any labeled datasets, allowing the model to learn directly from vast amounts of data. In addition to the usual challenges related to choice of model architecture and loss functions that enable good generalization, this project presented new infrastructure challenges due to the size of the models and datasets, which we successfully tackled.

World models are particularly valuable in autonomous driving because they not only simulate environmental dynamics and predict the motion of other vehicles and objects but also learn about semantic information, which enhances both planning and control within the autonomous driving stack. Being at the forefront of this cutting-edge research is both exciting and rewarding.

4. What skills do you find most essential in your role, and how have you developed these skills over time?

In my role, critical thinking is crucial. It’s important to question every experiment result, whether it’s worse or better than anticipated, as this helps deepen our understanding of underlying mechanisms. To develop these skills, I regularly read relevant research papers, stay updated on the latest advancements, and engage in discussions with fellow researchers. These conversations offer valuable insights into the challenges others have faced and the solutions they’ve found, which has greatly enhanced my problem-solving abilities.

5. What’s something about Plus that people might be surprised to learn?

The company was founded by a group of serial entrepreneurs, so there’s a heavy emphasis on results and making sure that what we’re building solves real-world problems for customers. That’s why we have a product portfolio that covers all levels of autonomy to meet different use cases, and we are partners to the world’s largest OEMs and Tier 1 suppliers like Bosch, Hyundai, Iveco, Traton, and others. This is made possible by our team’s exceptional AI talent, practical strategy for commercialization, and relentless execution.

6. What do you enjoy doing in your downtime?

I enjoy hiking and have visited most of the national parks in the Bay Area. I also take solo trips to national parks across the country to immerse myself in nature.

Besides hiking, I like playing badminton and watching procedural crime TV shows to unwind and relax.